SlideShare une entreprise Scribd logo
1  sur  19
Télécharger pour lire hors ligne
The current issue and full text archive of this journal is available at
                                         www.emeraldinsight.com/1463-5771.htm




                                                                                                                      Indian business
  Performance modeling of Indian                                                                                              schools
  business schools: a DEA-neural
        network approach
                                                                                                                                            221
                                          S. Sreekumar
         Rourkela Institute of Management Studies, Rourkela, India, and
                                        S.S. Mahapatra
   Department of Mechanical Engineering, National Institute of Technology,
                             Rourkela, India

Abstract
Purpose – The main purpose of the present study is to develop an integrated approach combining
data envelopment analysis (DEA) and neural network (NN) for assessment and prediction of
performance of Indian B-schools for effective decision making as error and biasness due to human
intervention in decision making is appreciably reduced.
Design/methodology/approach – DEA, being a robust mathematical tool, has been employed to
evaluate the efficiency of B-schools. DEA, basically, takes into account the input and output
components of a decision-making unit (DMU) to calculate technical efficiency (TE). TE is treated as an
indicator for performance of DMUs and comparison has been made among them. A sensitivity
analysis has been carried out to study robustness of the ranking of schools obtained through DEA.
Finally, NN is used to predict the efficiency when changes in inputs are caused due to market
dynamism so that effective strategies can be evolved by the managers with limited available data.
Findings – A total of 49 Indian B-schools are chosen for benchmarking purpose. The average score of
efficiency is 0.625 with a standard deviation of 0.175 when Charnes, Cooper and Rhodes (CCR) model is
used. Similarly, when the Banker, Charnes and Cooper (BCC) model is used the average score is 0.888 with a
standard deviation of 0.063. The rank order correlation coefficient between the efficiency ranking obtained
through CCR and BCC model is 0.736 ( p ¼ 0.000) which is significant. The peer group and peer weights for
the inefficient B-schools have been identified. This is useful for benchmarking for the inefficient DMUs.
They can identify the parameters in which they lack and take necessary steps for improvement. The peer
group for the inefficient B-schools indicates the efficient B-schools to which the inefficient B-schools are
closer in its combination of inputs and outputs. The TE obtained through DEA is used as output variable
along with input variables considered in DEA as input and output parameters in a generalized regression
NN during training phase. It can be observed that root mean square error is 0.009344 and 0.02323 for CCR-
and BCC-efficiency prediction, respectively, during training. Similarly, root mean square error is 0.08585
and 0.03279 for CCR- and BCC-efficiency prediction, respectively, during testing. Now, individual schools
can generate scenario with the data within their control and test their own performance through NN model.
Originality/value – This work proposes integration of DEA and NN to assist the managers to
predict the performance of an individual DMU based on input consumed and generate various
“what-if” scenarios. The study provides a simple but comprehensive methodology for improving
performance of B-schools in India.
Keywords Benchmarking, Data analysis, Decision making units, Neural nets
Paper type Research paper
                                                                                                                      Benchmarking: An International
                                                                                                                                               Journal
                                                                                                                                   Vol. 18 No. 2, 2011
1. Introduction                                                                                                                            pp. 221-239
India has liberalized the business education market in 1990s resulting in a rapid                                  q Emerald Group Publishing Limited
                                                                                                                                            1463-5771
growth of business schools offering programs at both undergraduate as well as                                         DOI 10.1108/14635771111121685
BIJ    post-graduate levels. The development of B-schools is largely adopting the policy of
18,2   self-sustainability and maximum of them are self-funded operated by private promoters.
       Most of the recruiters in India consider qualification in management as an added
       advantage. It caused demand in management education leading to an intense
       competition among the B-schools in the country. The low investment for entry and
       flourishing market has engineered the growth of B-schools throughout the country.
222    Mintzberg (1973) has pointed that the management school gives students degrees but it
       hardly teach them how to manage. Therefore, such degrees can barely be considered as
       prerequisites for managing firms in professional manner. On the contrary, Whitley et al.
       (1981) have advocated that many employers perceive holders of business education
       degree obviously distinguishes from those who do not possess it. Generally, students
       feel that getting a management degree from a reputed school may act as a formal way to
       batter career planning. Indian B-schools play a major role for providing career
       opportunity for around 68 percent of the Indian population who are in the 22-27 years age
       group. The quality of education imparted in Indian B-schools is reasonably good enough
       and many firms in the globe prefer Indian management graduates as a result of
       globalization and liberalization of the market economy. As a business strategy,
       manpower of cross-cultural nature may have edge over competitors (Sahay and Thakur,
       2007). Dayal (2002) emphasized on changing the structure of management education in
       India and suggested a strategy for institutional development for upgrading the quality
       of the academic program. The Indian B-schools should understand the emerging context
       of the economy, the industries, business and their needs and work out what they are
       delivering today and what they are expected to deliver tomorrow (Sahay and Thakur,
       2008). In this context, it is important for each of the Indian B-schools to know where one
       stands and design programs and pedagogy which will meet the future business needs.
       The new B-schools can set well-established institutes as their peers and follow them to
       become competitive. Although there is an unprecedented growth of schools in recent
       times, assessment on performance and efficiency of them is found to a limited extent in
       the literature. Measuring efficiency levels of the B-schools is an important issue for
       prospective students, parents, employers, and program administrators. Ranking of them
       can provide useful guidelines to all the stake holders involved in management education
       (Ojha, 2005). Some magazines like Outlook, Business World, Indian Management, etc.
       publish the annual report on the ranks of the Indian B-schools. However, ranking
       mechanism and sample size is questionable in such efforts.
           In this paper, a non-parametric technique called data envelopment analysis (DEA) is
       adopted to rank Indian B-schools based on their efficiency score. The scores can suggest
       inefficient and low-performing schools in an effective manner. Though the concept of
       benchmarking is good for improving the performance of individual unit, the problem
       associated with it is lack of transparency in data sharing. Therefore, methodology which
       will allow the individual schools to generate scenario with the data within their control
       and perform at the desired level is highly desirable. To address this problem, a neural
       network (NN) model is developed and trained to predict the performance level of the
       individual B-schools. The proposed model, in this paper, integrates the DEA and NN
       models to predict the performance of Indian B-schools. The data necessary for this study
       are collected from standard weekly business magazines and journals. The data sources
       employ professional surveying agencies for data collection. Therefore, data are
       considered to be reliable although collected from secondary sources. For the purpose
of the confidentiality of the schools and avoidance of conflicting interests, the identities     Indian business
of schools are not disclosed. However, some of the top Indian B-schools such as Indian                 schools
Institute of Management, Xavier Labour Relations Institute, Management Development
Institute, and Faculty of Management Studies (Delhi University) have been included in
the dataset.

2. Literature review                                                                                     223
In the recent years, several studies have been undertaken for analysis of efficiency in
education sector using DEA methodology. Each study differs in its scope, meaning, and
definition of decision-making units (DMUs). Tomkins and Green (1988) conducted DEA
analysis to test the performance of 20 accounting departments in UK. Johnes and Johnes
(1993) investigated the use of DEA in the assessment of performance of university
departments of the UK over the period 1984-1988. McMullen (1997) has applied DEA to
assess the relative desirability of Association to Advanced Collegiate Schools of
Business-accredited MBA programs. The authors have incorporated several attributes of
MBA programs into the model for finding out most desirable program in terms of these
attributes. McMillan and Datta (1998) have assessed the relative efficiency of 45 Canadian
universities using DEA. A subset of universities including universities from each of three
categories such as comprehensive with medical school, comprehensive without medical
school, and primarily undergraduate are regularly found efficient while some universities
exhibit inefficiency. But, overall efficiency for most of the universities is relatively high.
Ramanathan (2001) has compared the performance of selected schools in The Netherlands
using DEA and found that the efficiencies of the schools are closely related with their
performance. The authors have also observed that several non-discretionary input
variables can influence the efficiency scores but some of them are not in direct control of
management of the school. Lopes and Lanzer (2002) have addressed the issue of
performance evaluation, productivity, and quality of academic departments at a university
using a DEA model for cross-evaluation between departments considering the dimensions
of teaching, research, and service quality. The authors have observed zero correlation
between department teaching, research, and service and weak correlation between research
productivity and quality. Ray and Jeon (2003) in their study employed a measure of
Pareto-Koopmans global efficiency to evaluate the efficiency levels of MBA programs in
Business Week’s top-rated list. They computed input and output-oriented radial and
non-radial efficiency measures for comparison purpose. Among three-tier groups, the
schools from a higher tier group on an average are more efficient than those from lower tiers
although variations in efficiency levels do occur within the same tier. In India,
comparatively less studies have been conducted for performance evaluation of B-schools
using DEA. Wadhwa et al. (2005) has proposed integration of DEA and knowledge
management methods to evaluate the efficiency of technical education system (TES) in
India. The authors claim that the suggested approach can assist decision makers in
selecting proper institutes to further strengthen the TES in an efficient and effective
manner. A number of successful business applications of artificial neural networks (ANNs)
have been discussed in the literature, particularly in financial services (Tam and Kiang,
1992), transportation services (Nordmann and Luxhoj, 2000), telecommunications
(Mozer et al., 1999), etc. Lu et al. (1996) have compared the effectiveness of NNs and the
multinomial logit model, and concluded that the ANNs perform better than logit regressions
in franchising decision making. Wu et al. (1995) have applied NN approach for the decision
BIJ    surface modeling of apparel retail operations. Tam and Kiang (1992) have discussed a back
18,2   propagation NN application in predicting bankruptcy of financial institutions based on
       financial ratios. Dutta and Shekhar (1988) have applied NNs to a generalization problem of
       predicting the corporate bond ratings. Chiang et al. (1996) have discussed a back
       propagation NN approach to mutual fund net asset value forecasting. Hu et al. (2004) have
       found that ANN can perform better than logistic regression in the modeling of foreign
224    equities. Kimoto et al. (1990) have applied modular NNs to develop a buying and selling
       timing prediction system for stocks on the Tokyo Stock Exchange using a high-speed
       learning method called supplementary learning. Odom and Sharda (1990) have developed
       an NN model using back propagation for prediction of bankruptcy and compared results
       with discriminant analysis. It is claimed that ANN model performs better than discriminant
       analysis which is generally used for such type of problems. In education sector, some ANN
       models have been reported for prediction of academic performance of educational
       institutions considering qualitative as well as quantitative criteria (Hoefer and Gould, 2000;
       Kannan, 2005; Naik and Ragothaman, 2004; Wang, 1994). However, the application of NNs
       to model qualitative and intangible aspects of different services is not addressed adequately
       in the literature of education. It may be apposite to extend implementation of NNs to address
       more general and theoretical issues in service sector, such as education.

       3. Methodology
       3.1 Data envelopment analysis
       DEA, introduced by Charnes et al. (1978), computes efficiency score of each unit by
       comparing the efficiency score of each unit with that of its peers. Geometrically, a
       frontier can be constructed comprising of best performers. The units lying on the
       frontier are said to be efficient, and other units are treated as inefficient.
          Algebraically, the DEA model can be written as:
                      Pn                        Xn            Xm
                            ur yrj0
          max hj0 ¼ Pr¼1 m           subject to     ur yrj0 2     vi xij0 # 0 ur ; vi $ 0 ;r; i ð1Þ
                         i¼1 vi xij0            r¼1           i¼1

       where:
            hj0 ¼ relative efficiency of target DMU j0.
            r ¼ 1, 2,. . .n the number of outputs.
            i ¼ 1, 2,. . .m the number of inputs.
            j ¼ 1, 2,. . .s the number of DMUs.
            ur ¼ weight attached to the output r.
            vi ¼ weight attached to the input i.
            yrjo ¼ quantity of rth output produced by the DMU Jo.
            xijo ¼ quantity of ith input consumed by the DMU Jo.
       The DEA models may have any of the two orientations viz. input orientation and
       output orientation. Input orientation means how much inputs can be reduced while
       maintaining the same level of output. But output orientation of DEA is characterized
       by how much output can be increased while keeping the level of inputs constant.
       The latter orientation is more relevant for many service providers where the objective
       is to maximize the output maintaining the same level of inputs.
Another variation to a DEA model is the returns to scale (RTS) assumption. Constant,     Indian business
decreasing, increasing, and variable RTS assumptions may be employed. Constant return
to scale (CRS) implies that doubling inputs will exactly double outputs. Decreasing return
                                                                                                     schools
to scale implies that doubling inputs will less-than-double outputs. Increasing return to
scale implies that doubling inputs will more-than-double outputs. Thus, variable return
to scale (VRS) allows for a combination of constant, increasing, and decreasing inputs
and outputs. The DEA model shown in Equation (1) assumes a CRS. The drawback with                      225
the CRS model is that it compares DMUs only based on overall efficiency assuming
constant RTS. It ignores the fact that different DMUs could be operating at different
scales. To overcome this drawback, Banker et al. (1984) developed a model which
considers variable RTS and compares DMUs purely on the basis of TE. The model can be
shown as below:
               min u
                             Xn
               subject to        li xji 2 uxjj0 # 0 ;j
                             i¼1
                 X n                                                                   ð2Þ
                      li yrj 2 yjj0 $ 0 ;r
                 i¼1
                li ¼ 1 ;i
where:
     u ¼ efficiency score.
     li ¼ dual variable.
The difference between the CRS model (1) and the VRS model (2) is that the li is
restricted to one. This has the effect of removing the constraint in the CRS model that
DMUs must be scale efficient. Consequently, the VRS model allows variable RTS and
measures only TE for each DMU. Thus, a DMU to be considered as CRS efficient,
it must be both scale and technical efficient. For a DMU to be considered VRS efficient,
it only needs to be technically efficient.

3.2 Neural network
An ANN is an information processing paradigm that is inspired by the way biological
nervous systems, such as the human brain and process information. It is composed of a
large number of highly interconnected processing elements (neurons) working in
conjunction to solve specific problems. ANNs, like people, learn by example. An ANN is
configured for a specific application such as pattern recognition or data classification
through a learning process. Learning in biological systems involves adjustments to the
connections that exist between the neurons which is true for ANNs as well. An NN
consists of a network of neurons. Each neuron is associated with an input vector,
a weight vector corresponding to the input vector, a scalar bias, a transfer function, and
an output vector as shown in Figure 1. An NN may consist of one or more neurons in each
layer. In a network, the final layer is called the output layer and all previous layers are
called hidden layers. In the hidden layers, the output of a layer becomes the input for
the following layer. The transfer function of a neuron converts the input to the output of
the neuron. Multi-layer NNs are quite powerful tools used in solving many different
complex problems. Various types of NNs are available for different purposes. In this
study, a multi-layer back propagation NN architecture is adopted.
BIJ                                                                            Teach/use
18,2                                                        W1
                                           X
                                                            W2
                                           X

                                        Inputs            Weights
226                                                                             Neuron              Output
                                                           Wn
                                           X


Figure 1.
A typical neuron
                                                                             Teaching input


                           A typical NN is shown in Figure 2. There are three layers – a layer of “input” units is
                           connected to a layer of “hidden” units, which is connected to a layer of “output” units.
                           The behavior of the output units depends on the activity of the hidden units and the
                           weights between the hidden and output units. The architectures of ANN may be single
                           layer or multi-layer. In the single-layer organization, all units are connected to one
                           another. It constitutes the most general case and is of more potential computational
                           power than hierarchically structured multi-layer organizations.
                              As discussed, NNs are capable of learning complex relationships in data. The problems
                           NNs are used for can be divided in two general groups: classification problems in which
                           one tries to determine what type of category an unknown item falls into and numeric
                           problems where one attempts to predict a specific numeric outcome (Palisade Corporation,
                           2008). There are many computer software packages available for building and analysing
                           NNs. In this work, Neural Tools Version 5.0 by Palisade Corporation (2008) is used.
                           This software automatically scales all input data. Scaling involves mapping each variable
                           to a range with minimum and maximum values of 0 and 1. A non-linear scaling function
                           known as “tanh” is used as activation function. This function tends to squeeze data

                                                  Input             Hidden                 Output



                                        Input 1



                                        Input 2
                                                                                                    Output

                                        Input 3


Figure 2.
A simple feed forward NN                Input 4
with three layers
together at the low and high ends of the original data range (Mostafa, 2009). An NN            Indian business
configuration called generalized regression neural networks (GRNN) put forward                          schools
by Specht (1991) is adopted to give the best possible predictions. The rationale for
choosing the GRNN configuration lies in the fact that it is a good numerical predictor and
user need not to make decisions about the structure of a net. These nets always have two
hidden layers of neurons, with one neuron per training case in the first hidden layer, and
the size of the second layer determined by some facts about training data.                                 227
   GRN architecture. A generalized regression neural (GRN) net for two independent
numeric variables is structured as shown in Figure 3 with the assumption that there
are just three training cases (Palisade Corporation, 2008).
   The pattern layer contains one node for each training case. Presenting a training case to
the net consists here of presenting two independent numeric values. Each neuron in the
pattern layer computes its distance from the presented case. The values passed to the
numerator and denominator nodes are functions of the distance and the dependent value.
The two nodes in the summation layer sum its inputs, while the output node divides them
to generate the prediction. The distance function computed in the pattern layer neurons
uses “smoothing factors”; every input has its own “smoothing factor” value. With a single
input, the greater the value of the smoothing factor, the more significant distant training
cases become for the predicted value. With two inputs, the smoothing factor relates to the
distance along one axis on a plane, and in general, with multiple inputs, to one dimension
in multi-dimensional space. Training a GRN net consists of optimizing smoothing factors
to minimize the error on the training set, and the conjugate gradient descent optimization
method is used to accomplish that. The error measure used during training to evaluate
different sets of smoothing factors is the mean squared error. However, when computing
the squared error for a training case, that case is temporarily excluded from the pattern
layer. This is because the excluded neuron would compute a zero distance, making other
neurons insignificant in the computation of the prediction.

3.3 Data and data classification
For solving the benchmarking problem, 49 top B-schools of India are considered using
convenience sampling method. Data on 11 parameters as listed below are collected from
various secondary sources. The secondary source reference includes popular Indian
magazines like Outlook, Business World, Indian Management, and B-school directories
which publish the annual report on the ranks of the Indian B-schools (Table I).




                                                                      Output


                                                                                                        Figure 3.
                                                                                                    A simple GRN
               Inputs            Pattern            Summation                                         architecture
                                  later               layer
BIJ                  The data collected on the above parameters are classified into two categories based on
18,2                 their nature for DEA and NN application. The criteria of selection of inputs and
                     outputs are quite subjective; there is no specific rule for determining the procedure for
                     selection of inputs and outputs (Ramanathan, 2001). The classification of input and
                     output is done as follows (Sreekumar and Patel, 2007):
                            Input:
228                            X1: IC
                               X2: IF
                               X3: FEE
                            Output:
                               Y1: II
                               Y2: PP
                               Y3: IL
                               Y4: RS
                               Y5: SS
                               Y6: FS
                               Y7: ECA
                               Y8: SAL


                     S. no. Parameter              Abbreviation Explanation

                      1    Intellectual capital    IC           Faculty/student ratio, teaching experience of faculty,
                                                                corporate experience of faculty/students, PhD/students
                                                                ratio, faculty with PhD (abroad), books, research papers,
                                                                and cases
                      2    Industry interface      II           Revenue from consultancy, revenue from management
                                                                development programs, seminars, and workshops
                      3    Infrastructure and      IF           Area (in acres), built-up area, computers per batch,
                           facilities                           amphitheatre class room, library books, electronic
                                                                database, residential facilities, single occupancy room,
                                                                and MDP hostel
                      4    International linkage   IL           Student exchange program and faculty exchange
                                                                program
                      5    Placement               PP           Percentage of student placed, median salary, maximum
                           performance                          salary, minimum salary, percentage of students placed
                                                                abroad, and return on investment
                      6    Extra curricular        ECA          National-level events organized and awards won
                           activities                           by students
                      7    Recruiters              RS           Application of knowledge of subject/skills,
                           satisfaction                         analytical skills, communication and presentation
                                                                skills, creativity, proactive attitude, and ability to
                                                                work in team
                      8    Students satisfaction   SS           Satisfaction of ongoing students from the school
                      9    Faculty satisfaction    FS           Based on present faculty of the school
Table I.             10    Fee                     FEE          Fee collected from students
List of parameters   11    Salary                  SAL          Initial salary at which graduating students are placed
In this study, the DEA and NN has been integrated to have an efficient predicting model.       Indian business
DEA, being a robust mathematical tool, has been employed to evaluate the efficiency of                 schools
B-schools. DEA, basically, takes into account the input and output components of a
DMU to calculate TE. TE is treated as an indicator for performance of DMUs and
comparison has been made among them. A sensitivity analysis has been carried out to
study robustness of the ranking of schools obtained through DEA. Finally, NN is used
to predict the efficiency when changes in inputs are caused due to market dynamism so                    229
that effective strategies can be evolved by the managers with limited available data.

4. Results and discussion
The Charnes, Cooper and Rhodes (CCR)-DEA model as discussed above is based on
constant RTS does not consider the size of B-school under consideration while
calculating the efficiency. But in many cases the size of a unit may influence its ability to
produce services more efficiently. So, we have also considered the VRS model for our
study. The B-schools under consideration for our problem contain both private and
government institute. The input for both the category of institute differs widely, so the
output orientation model is used. It may be noted that the Banker, Charnes and Cooper
(BCC) model allows variable RTS and measures only TE for each DMU whereas a DMU
is considered as CCR efficient if it is both scale and technical efficient.
    The relative efficiency score of B-schools are analysed and presented in Table II. The
BCC score is based on VRS assumption and measures the pure TE. The CCR score
is based on CRS assumption and consist of non-additive combination of pure TE and
scale efficiency. The table shows that in a scale of 0-1 the average score for the B-schools
is 0.625 with a standard deviation of 0.175 when CCR model is used. Similarly, when the
BCC model is used the average score is 0.888 with a standard deviation of 0.063. The
rank-order correlation coefficient between the efficiency ranking obtained through CCR
and BCC model is 0.736 ( p ¼ 0.000) which is significant. The above table also shows the
peer group and peer weights for the inefficient B-schools. This is useful for
benchmarking for the inefficient DMUs. They can identify the parameters in which they
lack and take necessary steps for improvement. The peer group for the inefficient
B-schools indicates the efficient B-schools to which the inefficient B-schools are closer in
its combination of inputs and outputs. It may also be observed that in both CCR and BCC
score there are multiple numbers of DMUs with efficiency score unity leading to tie case.
The school which appears maximum number of times as peer in the above table may be
treated as the best school. Moreover, this school is likely to be the school which is
efficient with respect to a large number of factors, and is probably a good example of an
exemplary operating performer. Efficient DMUs that appear seldom in the peer set of
other inefficient DMUs are likely to possess a very uncommon input/output mix and are
thus not suitable examples for other inefficient schools (Mostafa, 2009). Now, it is
prudent to check the robustness of the model trough sensitivity.
    DEA is an extreme point technique because the efficiency frontier is formed by the
actual performance of best-performing DMUs. A direct consequence of this aspect is that
errors in measurement can affect the DEA result significantly. So, according to DEA
technique, it is possible for a B-School to become efficient if it achieves exceptionally
better results in terms of one output but performs below average in other outputs. The
sensitivity of DEA efficiency can be verified by checking whether the efficiency of a
DMU is affected appreciably:
BIJ
                                                                                                                                                          18,2


                                                                                                                                            230




 oriented)
 Table II.
 Efficiency score (output
DMU                         CCR       Rank   Peer                Peer weights            BCC       Rank     Peer                          Peer weights

 1                         1.000000     1              1                –               1.000000     1           1, 1                             –
 2                         1.000000     1         2,   1                –               1.000000     1           2, 1                             –
 3                         0.995190     5    1,   2,   6   0.634, 0.382, 2.71 £ 102 2   1.000000     1           3, 1                             –
 4                         1.000000     1         4,   1                –               1.000000     1           4, 1                             –
 5                         0.877809     7         1,   4   1.051, 0.306                 1.000000     1           5, 1                             –
 6                         1.000000     1         6,   1                –               1.000000     1           6, 1                             –
 7                         0.799198     9    2,   4,   6   0.333, 0.674, 0.455          1.000000     1           7, 1                             –
 8                         0.844716     8    2,   4,   6   0.341, 0.548, 0.296          0.964907     9       2, 4, 17   0.570,   0.330,   0.101
 9                         0.618562    20    1,   2,   6   0.159, 0.631, 0.741          0.907631    16     1, 2, 5, 6   0.111,   0.429,   0.177, 0.283
10                         0.644462    19         4,   6   1.050, 0.491                 0.898023    19        1, 2, 7   0.150,   0.508,   0.342
11                         0.524947    31    2,   4,   6   0.704, 0.804, 0.333          0.902703    18        2, 5, 6   0.681,   0.289,   0.030
12                         0.753229    10         2,   6   1.209, 1.11 £ 102 2          0.907649    15           2, 7   0.884,   0.115
13                         0.584787    23         4,   6   1.141, 0.471                 0.881374    21        1, 2, 7   0.487,   0.393,   0.120
14                         0.729801    13         4,   6   0.659, 0.687                 0.943011    11        2, 4, 6   0.486,   0.215,   0.299
15                         0.514123    32         4,   6   1.147, 0.653                 0.869822    25              2   1.000
16                         0.526763    29    2,   4,   6   0.341, 1.090, 0.135          0.793427    47           1, 7   0.819,   0.181
17                         0.904431     6         4,   6   1.00 £ 102 1, 1.079          1.000000     1             17   1.000
18                         0.546956    26         4,   6   0.614, 1.227                 0.905501    17       4, 6, 17   0.670,   0.226, 0.104
19                         0.568486    24         4,   6   0.621, 1.141                 0.921440    13    4, 6, 7, 17   0.366,   0.286, 0.301, 4.79 £ 102 2
20                         0.508409    33         4,   6   0.353, 1.579                 0.912566    14           6, 7   0.586,   0.414
21                         0.556489    25         4,   6   0.634, 0.969                 0.861150    28           2, 6   0.703,   0.297
22                         0.504688    34         4,   6   0.905, 1.010                 0.872109    24        2, 6, 7   0.692,   1.07 £ 102 2, 0.297
23                         0.534655    27         4,   6   0.208, 1.415                 0.856631    33           2, 6   0.266,   0.734
24                         0.679945    15         4,   6   0.580, 0.744                 0.857588    31           2, 6   0.624,   0.376
25                         0.470277    44         4,   6   0.613, 1.356                 0.843978    39        2, 4, 6   0.571,   0.136, 0.292
26                         0.485306    39         4,   6   0.605, 1.349                 0.869303    26     2, 4, 6, 7   0.533,   6.42 £ 102 2, 0.305, 9.74 £ 102 2
27                         0.484885    40         4,   6   0.6135, 1.239                0.831249    41           2, 6   0.703,   0.297
28                         0.472438    43         4,   6   1.044, 1.141                 0.881026    22           2, 7   0.131,   0.869
29                         0.500204    36         4,   6   0.701, 1.119                 0.876344    23           2, 6   0.790,   0.210
30                         0.487138    37         4,   6   0.567, 1.342                 0.854651    34     2, 4, 6, 7   0.199,   0.238, 0.361, 0.202
31                         0.476137    41         4,   6   0.353, 1.572                 0.860350    30        2, 6, 7   0.110,   0.579, 0.311
32                         0.475426    42         4,   6   0.479, 1.460                 0.853220    37        2, 4, 6   0.453,   0.115, 0.431
                                                                                                                                                       (continued)
DMU            CCR       Rank    Peer            Peer weights           BCC        Rank        Peer                     Peer weights

 33           0.486999    38      4,   6   0.444, 1.408               0.860563       29            2, 6    0.528, 0.472
 34           0.525993    30      4,   6   0.744, 0.967               0.857388       32            2, 6    0.825, 0.175
 35           0.665816    17      4,   6   9.41 £ 102 3, 1.453        0.945689       10        6, 7, 17    0.703, 1.38 £ 102 2, 0.283
 36           0.613720    22      4,   6   0.313, 1.064               0.829688       42            2, 6    0.354, 0.6463
 37           0.457457    45      4,   6   0.835, 1.010               0.806735       45            2, 6    0.930, 6.99 £ 102 2
 38           0.533847    28      4,   6   0.272, 1.508               0.883930       20            2, 6    0.236, 0.358, 5.64 £ 102 2, 0.349
 39           0.617969    21      4,   6   0.224, 1.199               0.854069       36         2, 4, 6    0.079, 0.181, 0.739
 40           0.675052    16      4,   6   0.488, 0.814               0.854545       35            2, 6    0.528, 0.472
 41           0.435924    47      4,   6   0.260, 1.778               0.838353       40         2, 4, 6    0.302, 5.05 £ 102 2, 0.648
 42           0.452255    46      4,   6   0.403, 1.446               0.813703       44            2, 6    0.485, 0.515
 43           0.739848    11      4,   6   9.72 £ 102 2, 1.221        0.938403       12        6, 7, 17    0.460, 6.65 £ 102 2, 0.473
 44           0.500706    35      4,   6   0.373, 1.315               0.827064       43            2, 6    0.441, 0.559
 45           0.695122    14           4   1.263                      0.866262       27        4, 6, 17    8.72 £ 102 3, 0.882, 0.109
 46           0.738727    12      4,   6   0.244, 0.926               0.848324       38        4, 6, 17    0.250, 0.677, 0.073
 47           0.355991    49      4,   6   0.969, 1.364               0.792899       48               2    1.000
 48           0.662169    18      4,   6   0.071636, 1.143831         0.800977       46            2, 6    9.17 £ 102 2, 0.908
 49           0.412507    48      4,   6   0.776855, 1.205363         0.784348       49               2    0.882, 0.118
Notes: Avg (CCR) – 0.625297, Min – 0.355991, and SD – 0.175012; Avg (BCC) – 0.888339, Min – 0.784348, and SD – 0.062977; Pearson correlation of
CCR and BCC ¼ 0.736, p-value ¼ 0.000
                                                                                                                                           Indian business
                                                                                                                                                   schools


                                                                                                                           231




  Table II.
BIJ       .
              if only one input or output is omitted from DEA analysis; and
18,2      .
              dropping one efficient DMU at a time from DEA analysis.

       Initially the input “intellectual capital” is dropped from the analysis and TE of DMUs is
       calculated, then input “FEE” is dropped, and similarly the outputs “placement performance”
       is dropped from both CCR and BCC model. At the second level, the efficient unit DMU1 is
232    dropped to calculate the CCR and BCC efficiency. The results of both the stage are tabulated
       in Table III. The table shows that dropping the input “IC” and outputs “PP” one-by-one
       causes no significant change in the TE score of DMUs and efficient units are remaining
       efficient as such. Change in efficiency score is observed when the input “FEE” is dropped
       from the analysis. DMU6 is becoming inefficient when “FEE” is not considered for CCR
       efficiency. This indicates that “FEE” is an important input for such schools. At the second
       level of analysis, some of the efficient DMUs are dropped one-by-one. It is observed that the
       efficient units are remaining efficient as such when DMU1 is dropped from the DEA
       analysis but DMU3 becomes an efficient unit for CCR efficiency whereas there is no change
       efficiency status for BCC score.
           Finally, GRNN model is used to predict efficiency score of DMUs. The prediction
       results help the managers to use the available data for strategic decision making when
       the data for benchmarking is not shared by DMUs under consideration. The prediction
       can also generate various scenarios to guide the managers/administrators for effective
       decision making. In addition, human error is avoided during prediction process. Both
       CCR- and BCC-efficiency scores are predicted using GRNN model. The three inputs
       considered for prediction purpose are IC, IF, and FEE. As mentioned earlier, Neural
       Tools version 5.0 (Palisade Corporation, 2008) is used for building the network due to its
       flexibility and extensive capabilities. In general, 75 percent of the data is considered for
       training for mapping good relationship between inputs and outputs and 25 percent is
       used for testing. In this study, 39 cases are used for training the network chosen
       randomly from dataset to provide variation during training and ten cases for testing.
       The model is used to predict the output for the entire dataset of 49 B-schools. The
       network configuration setting is shown in Table IV. The network is run till a reasonable
       root mean square error is attained during training period. The number of epochs to
       obtain error within tolerance limit happens to be 69 and 98, respectively, for CCR- and
       BCC-efficiency prediction during training of the network. Once the network is well
       trained, it can be used for testing purpose as generalization capability of the network is
       ensured. It can be observed that root mean square error is 0.009344 and 0.02323 for
       CCR- and BCC-efficiency prediction, respectively, during training. Similarly, root mean
       square error is 0.08585 and 0.03279 for CCR- and BCC-efficiency prediction, respectively,
       during testing. The prediction results are shown in Table V. It can be observed that
       maximum absolute difference in CCR-efficiency score and the NN prediction is 0.108369
       which occurs for DMU9. Similarly, maximum absolute difference in BCC-efficiency
       score and the NN prediction is 0.067631 which occurs again for DMU9. For rest of the
       cases, the absolute difference between either for CCR or BCC and neural network
       prediction is much smaller. The NN prediction of efficiency scores is compared with
       scores obtained through DEA using CCR and BCC in Figure 4. It is noted that the pattern
       is perfectly followed in both type of prediction.
Efficiency Efficiency     Dropping   Dropping   Dropping   Dropping   Dropping DMU Dropping DMU   Dropping   Dropping
DMU            CCR       BCC          IC, CCR    IC, BCC   FEE, CCR   FEE, BCC       1, CCR       1, BCC     PP, CCR    PP, BCC

 1             1.000000   1.000000   1.000000   1.000000   1.000000   1.000000         –            –        1.000000    1.000000
 2             1.000000   1.000000   1.000000   1.000000   1.000000   1.000000     1.000000     1.000000     1.000000    1.000000
 3             0.995190   1.000000   0.981913   1.000000   0.970112   1.000000     1.000000     1.000000     0.995190    1.000000
 4             1.000000   1.000000   1.000000   1.000000   1.000000   1.000000     1.000000     1.000000     1.000000    1.000000
 5             0.877809   1.000000   0.772174   1.000000   0.877809   1.000000     0.981736     1.000000     0.877809    1.000000
 6             1.000000   1.000000   1.000000   1.000000   0.586713   1.000000     1.000000     1.000000     1.000000    1.000000
 7             0.799198   1.000000   0.774525   1.000000   0.607166   1.000000     0.799198     1.000000     0.799198    1.000000
 8             0.844716   0.964907   0.844716   0.964907   0.673546   0.937684     0.844716     0.964907     0.844716    0.964907
 9             0.618562   0.907631   0.618562   0.907631   0.392137   0.900923     0.620586     0.907877     0.618562    0.907631
10             0.644462   0.898023   0.644462   0.898023   0.433763   0.898023     0.644462     0.899259     0.644462    0.898023
11             0.524947   0.902703   0.492203   0.902703   0.430246   0.902703     0.524947     0.902703     0.524947    0.902703
12             0.753229   0.907649   0.522483   0.886576   0.738128   0.907649     0.753229     0.907649     0.753229    0.907649
13             0.584787   0.881374   0.584787   0.881374   0.436566   0.881374     0.584787     0.888147     0.584787    0.881374
14             0.729801   0.943011   0.729801   0.943011   0.423904   0.902508     0.729801     0.943011     0.729801    0.943011
15             0.514123   0.869822   0.514123   0.869822   0.376683   0.869822     0.514123     0.869822     0.514123    0.869822
16             0.526763   0.793427   0.526763   0.793427   0.434564   0.793427     0.526763     0.804730     0.526763    0.793427
17             0.904431   1.000000   0.904431   1.000000   0.355112   0.832466     0.904431     1.000000     0.904431    1.000000
18             0.546956   0.905501   0.546956   0.905501   0.262281   0.846514     0.546956     0.905501     0.546956    0.905501
19             0.568486   0.921440   0.568486   0.921440   0.320073   0.879471     0.568486     0.921440     0.568486    0.921440
20             0.508409   0.912566   0.508409   0.912566   0.274155   0.867030     0.508409     0.912566     0.508409    0.912566
21             0.556489   0.861150   0.556489   0.861150   0.331672   0.852071     0.556489     0.861150     0.556489    0.861150
22             0.504688   0.872109   0.504688   0.872109   0.324421   0.871695     0.504688     0.872109     0.504688    0.872109
23             0.534655   0.856631   0.534655   0.856631   0.265378   0.834320     0.534655     0.856631     0.534655    0.856631
24             0.679945   0.857588   0.679945   0.857588   0.359272   0.846154     0.679945     0.857588     0.679945    0.857588
25             0.470277   0.843978   0.470277   0.843978   0.249000   0.828402     0.470277     0.843978     0.470277    0.843978
26             0.485306   0.869303   0.485306   0.869303   0.270222   0.853810     0.485306     0.869303     0.485306    0.869303
27             0.484885   0.831249   0.484885   0.831249   0.257981   0.822485     0.484885     0.831249     0.484885    0.831249
28             0.472438   0.881026   0.472438   0.881026   0.311123   0.881026     0.472438     0.881026     0.472438    0.881026
29             0.500204   0.876344   0.500204   0.876344   0.273313   0.869822     0.500204     0.876344     0.500204    0.876344
30             0.487138   0.854651   0.487138   0.854651   0.260696   0.827372     0.487138     0.854651     0.487138    0.854651
31             0.476137   0.860350   0.476137   0.860350   0.249919   0.838258     0.476137     0.860350     0.476137    0.860350
                                                                                                                        (continued)




               report
 Sensitivity analysis
                                                                                                                           Indian business
                                                                                                                                   schools




          Table III.
                                                                                                               233
BIJ
                                                                                                                          18,2


                                                                                                              234




 Table III.
              Efficiency Efficiency   Dropping   Dropping   Dropping   Dropping   Dropping DMU Dropping DMU   Dropping   Dropping
DMU             CCR       BCC        IC, CCR    IC, BCC   FEE, CCR   FEE, BCC       1, CCR       1, BCC     PP, CCR    PP, BCC

32            0.475426   0.853220   0.475426   0.853220   0.238986   0.834320     0.475426     0.853220     0.475426   0.853220
33            0.486999   0.860563   0.486999   0.860563   0.239228   0.846154     0.486999     0.860563     0.486999   0.860563
34            0.525993   0.857388   0.525993   0.857388   0.288255   0.852071     0.525993     0.857388     0.525993   0.857388
35            0.665816   0.945689   0.665816   0.945689   0.301387   0.849922     0.665816     0.945689     0.665816   0.945689
36            0.613720   0.829688   0.613720   0.829688   0.289507   0.810651     0.613720     0.829688     0.613720   0.829688
37            0.457457   0.806735   0.457457   0.806735   0.257054   0.804734     0.457457     0.806735     0.457457   0.806735
38            0.533847   0.883930   0.533847   0.883930   0.255412   0.804734     0.533847     0.883930     0.533847   0.883930
39            0.617969   0.854069   0.617969   0.854069   0.283359   0.822485     0.617969     0.854069     0.617969   0.854069
40            0.675052   0.854545   0.675052   0.854545   0.339557   0.840237     0.675052     0.854545     0.675052   0.854545
41            0.435924   0.838353   0.435924   0.838353   0.205649   0.816568     0.435924     0.838353     0.435924   0.838353
42            0.452255   0.813703   0.452255   0.813703   0.230183   0.798817     0.452255     0.813703     0.452255   0.813703
43            0.739848   0.938403   0.739848   0.938403   0.286491   0.835148     0.739848     0.938403     0.739848   0.938403
44            0.500706   0.827064   0.500706   0.827064   0.254029   0.810651     0.500706     0.827064     0.500706   0.827064
45            0.695122   0.866262   0.695122   0.866262   0.208424   0.822485     0.695122     0.866262     0.695122   0.866262
46            0.738727   0.848324   0.738727   0.848324   0.318954   0.804734     0.738727     0.848324     0.738727   0.848324
47            0.355991   0.792899   0.355991   0.792899   0.205402   0.792899     0.355991     0.792899     0.355991   0.792899
48            0.662169   0.800977   0.662169   0.800977   0.275852   0.775148     0.662169     0.800977     0.662169   0.800977
49            0.412507   0.784348   0.412507   0.784348   0.237532   0.781065     0.412507     0.784348     0.412507   0.784348
Indian business
                                             CCR-efficiency prediction     BCC-efficiency prediction
                                                                                                               schools
Net information
Configuration                                 GRNN numeric predictor       GRNN numeric predictor
Independent category variables               0                            0
Independent numeric variables                3 (IC, IF, and FEE)          3 (IC, IF, and FEE)
Dependent variable                           Numeric variable (CCR-eff)   Numeric variable (BCC-eff)                 235
Training
Number of cases                                      39                           39
Number of trials                                     69                           98
Bad predictions (%) (30 percent tolerance)            0.0000                       0.0000
Root mean square error                                0.009344                     0.02323
Mean absolute error                                   0.005236                     0.01606
Std deviation of abs. error                           0.007740                     0.01679
Testing
Number of cases                                      10                           10                              Table IV.
Bad predictions (%) (30 percent tolerance)            0.0000                       0.0000              Network configuration
Root mean square error                                0.08585                      0.03279                 for CCR and BCC
Mean absolute error                                   0.05931                      0.02420                   efficiency score
Std deviation of absolute Error                       0.06208                      0.02212                        prediction



5. Conclusions
This study is concerned with the ranking of Indian B-schools using a non-parametric
technique and developing an NN to predict the standing of schools based on limited
number of parameters. Both CRS and VRS are considered to obtain the efficiency score
of DMUs. The methodology facilitates in identifying the benchmarked institutions for
the inefficient institutes. The process of benchmarking is useful in identifying the best
business practices and formulating the winning strategies. The DEA methodology can
be quite useful for Indian B-schools in identifying their position relative to their peers,
and in formulating strategies for improvement by right mix of inputs and outputs.
Although the concept of benchmarking is good for improving the performance of
individual unit, the problem associated with it is lack of transparency in data sharing
and data reliability. Some of the schools may not be too enthusiastic to share the data
pertaining to their resource consumption and output produced. This requires a
methodology which will allow the individual schools to generate scenario with the data
within their control and test their own performance through simulation. For this
purpose, an NN model is developed and trained to predict the performance level of the
individual B-schools based on the input level consumed. The paper integrates the DEA
model and NN. The output obtained through DEA model is used for training the NN
for prediction of efficiency score of schools based on the input values. Like any other
study, this paper also has several limitations giving opportunity for further research.
The paper considers 11 parameters relevant for improving quality of Indian B-schools.
The other pertinent factors like quality of inputs (students), investment pattern in the
institution, funds generation by the institution, etc. could have been incorporated
in the model for calculating efficiency score of schools. Next, some of the inputs may
not be fully under the control of management leading to practically infeasible target.
Again as the DEA gives the relative efficiency score, so it gets affected by sample size.
BIJ
                                            NN         Absolute                    NN        Absolute
18,2                    DMU   CCR-eff    prediction    difference    BCC-eff    prediction   difference

                         1    1.000000   0.995199     0.004801       1.000000   0.990000     0.010000
                         2    1.000000   0.999804     0.000196       1.000000   0.990000     0.010000
                         3    0.995190   0.995383     0.000190       1.000000   0.980000     0.020000
236                      4    1.000000   1.000000       9 £ 102 8    1.000000   0.970000     0.030000
                         5    0.877809   0.877809     8.5 £ 102 10   1.000000   1.000000     0.000000
                         6    1.000000   1.000000     2.76 £ 102 7   1.000000   1.000000     0.000000
                         7    0.799198   0.798370     0.000828       1.000000   0.990000     0.010000
                         8    0.844716   0.844631     8.5 £ 102 5    0.964907   0.970000     0.005090
                         9    0.618562   0.510193     0.108369       0.907631   0.840000     0.067631
                        10    0.644462   0.643716     0.000746       0.898023   0.890000     0.008023
                        11    0.524947   0.524947     1.41 £ 102 7   0.902703   0.920000     0.017300
                        12    0.753229   0.753229     1.4 £ 102 12   0.907649   0.950000     0.042350
                        13    0.584787   0.585286     0.000500       0.881374   0.880000     0.001374
                        14    0.729801   0.711399     0.018402       0.943011   0.880000     0.063011
                        15    0.514123   0.515267     0.001140       0.869822   0.880000     0.010180
                        16    0.526763   0.638785     0.112020       0.793427   0.800000     0.006570
                        17    0.904431   0.904431     3.26 £ 102 7   1.000000   0.980000     0.020000
                        18    0.546956   0.515410     0.031546       0.905501   0.860000     0.045501
                        19    0.568486   0.489394     0.079092       0.921440   0.910000     0.011440
                        20    0.508409   0.501404     0.007005       0.912566   0.860000     0.052566
                        21    0.556489   0.556487     2.26 £ 102 6   0.861150   0.860000     0.001150
                        22    0.504688   0.504462     0.000226       0.872109   0.890000     0.017890
                        23    0.534655   0.536646     0.001990       0.856631   0.870000     0.013370
                        24    0.679945   0.698285     0.018340       0.857588   0.870000     0.012410
                        25    0.470277   0.478689     0.008410       0.843978   0.860000     0.016020
                        26    0.485306   0.482410     0.002896       0.869303   0.860000     0.009303
                        27    0.484885   0.491726     0.006840       0.831249   0.860000     0.028750
                        28    0.472438   0.472462     2.4 £ 102 5    0.881026   0.880000     0.001026
                        29    0.500204   0.501399     0.001190       0.876344   0.870000     0.006344
                        30    0.487138   0.486487     0.000651       0.854651   0.850000     0.004651
                        31    0.476137   0.484822     0.008680       0.860350   0.840000     0.020350
                        32    0.475426   0.484047     0.008620       0.853220   0.850000     0.003220
                        33    0.486999   0.491297     0.004300       0.860563   0.860000     0.000563
                        34    0.525993   0.511250     0.014743       0.857388   0.860000     0.002610
                        35    0.665816   0.662770     0.003046       0.945689   0.950000     0.004310
                        36    0.613720   0.616240     0.002520       0.829688   0.840000     0.010310
                        37    0.457457   0.474462     0.017000       0.806735   0.840000     0.033270
                        38    0.533847   0.532308     0.001539       0.883930   0.880000     0.003930
                        39    0.617969   0.619263     0.001290       0.854069   0.840000     0.014069
                        40    0.675052   0.675232     0.000180       0.854545   0.850000     0.004545
                        41    0.435924   0.478165     0.042240       0.838353   0.840000     0.001650
                        42    0.452255   0.484692     0.032440       0.813703   0.850000     0.036300
                        43    0.739848   0.723855     0.015993       0.938403   0.880000     0.058403
                        44    0.500706   0.502156     0.001450       0.827064   0.840000     0.012940
                        45    0.695122   0.494601     0.200521       0.866262   0.860000     0.006262
                        46    0.738727   0.737289     0.001438       0.848324   0.870000     0.021680
Table V.                47    0.355991   0.355991     1.3 £ 102 8    0.792899   0.800000     0.007100
CCR and BCC efficiency   48    0.662169   0.674932     0.012760       0.800977   0.840000     0.039020
score prediction        49    0.412507   0.435469     0.022960       0.784348   0.840000     0.055650
1.2                                                                      Indian business
                               1
                                                                                                               schools

                              0.8                                         CCR-EFF
                 Efficiency



                              0.6                                         NN-CCR                                     237
                              0.4                                         BCC-EFF

                              0.2                                         NN-BCC
                                                                                                                  Figure 4.
                                                                                                          Comparison of NN
                               0                                                                            predictions with
                                    1 5 9 13 17 21 25 29 33 37 41 45 49
                                                                                                               DEA results
                                                 DMUs

In future study, more number of schools over a period of time may be considered for
better insight into the problem.

References
Banker, R.D., Charnes, A. and Cooper, W.W. (1984), “Some models for estimating technical and scale
      inefficiencies in data envelopment analysis”, Management Science, Vol. 30 No. 9, pp. 1078-92.
Charnes, A., Cooper, W.W. and Rhodes, E. (1978), “Measuring the efficiency of decision making
      units”, European Journal of Operations Research, Vol. 2 No. 6, pp. 429-44.
Chiang, W., Urban, T.L. and Baldridge, G.W. (1996), “A neural network approach to mutual fund
      net asset value forecasting”, Omega, Vol. 24 No. 2, pp. 205-15.
Dayal, I. (2002), “Developing management education in India”, Journal of Management Research,
      Vol. 2 No. 2, pp. 98-113.
Dutta, S. and Shekhar, S. (1988), “Bond ratings: a non-conservative application of neural networks”,
      IEEE International Conference on Neural Networks, San Diego, CA, Vol. 2, pp. 443-50.
Hoefer, P. and Gould, J. (2000), “Assessment of admission criteria for predicting students’
      academic performance in graduate business programs”, Journal of Education for Business,
      Vol. 75 No. 4, pp. 225-9.
Hu, M.Y., Zhang, G.P. and Haiyang, C. (2004), “Modeling foreign equity control in Sino-foreign
      joint ventures with neural networks”, European Journal of Operational Research, Vol. 159
      No. 3, pp. 729-40.
Johnes, G. and Johnes, J. (1993), “Measuring the research performance of UK economics
      departments: an application of data envelopment analysis”, Oxford Economics Papers,
      Vol. 4 No. 2, pp. 332-47.
Kannan, S.R. (2005), “Extended bidirectional associative memories: a study on poor education”,
      Mathematical and Computer Modelling, Vol. 42 Nos 3/4, pp. 389-95.
Kimoto, T., Asakawa, K., Yoda, M. and Takeoda, M. (1990), “Stock market prediction system
      with modular neural networks”, Proceedings of the International Joint Conference on
      Neural Networks, San Diego, CA, Vol. 1, pp. 1-6.
Lopes, A.L.M. and Lanzer, E.A. (2002), “Data envelopment analysis – DEA and fuzzy sets to
      assess the performance of academic departments: a case study at Federal University of
      Santa Catarina – UFSC”, Pesquisa Operational, Vol. 22 No. 2, pp. 217-30.
BIJ    Lu, L-C., Chen, W-H., Kim, D. and Hwang, C-P. (1996), “Artificial neural systems improve franchising
             decision making”, International Journal of Management, Vol. 13 No. 2, pp. 25-32.
18,2
       McMillan, L.M. and Datta, D. (1998), “The relative efficiencies of Canadian universities: a DEA
             perspective”, Canadian Public Policy, Vol. 24 No. 4, pp. 485-511.
       McMullen, P.R. (1997), “Assessment of MBA programs via data envelopment analysis”, Journal
             of Business and Management, Vol. 5 No. 1, pp. 77-91.
238    Mintzberg, H. (1973), The Nature of Managerial Work, Harper & Row, New York, NY.
       Mostafa, M.M. (2009), “Modeling the efficiency of top Arab banks: a DEA-neural network
             approach”, Expert Systems with Applications, Vol. 36, pp. 309-20.
       Mozer, M., Wolniewicz, R., Johnson, E. and Kaushansky, H. (1999), “Curn reduction in the wireless
             industry”, Proceedings of the Neural Information Systems Conference, San Diego, CA.
       Naik, B. and Ragothaman, S. (2004), “Using neural networks to predict MBA student success”,
             College Student Journal, Vol. 38, March, pp. 210-8.
       Nordmann, L.H. and Luxhoj, J.T. (2000), “Neural network forecasting of service problems for
             aircraft structural component grouping”, Journal of Aircraft, Vol. 37 No. 2, pp. 332-8.
       Odom, M.D. and Sharda, R. (1990), “A neural network model for bankruptcy prediction”,
             Proceedings of the International Joint Conference on Neural Networks, San Diego, CA,
             Vol. 2, pp. 163-8.
       Ojha, A.K. (2005), “Abhoy management education in India: protecting it from the rankings
             onslaught”, Decision, Vol. 32 No. 2, pp. 19-33.
       Palisade Corporation (2008), Neural Tools User Guide Version 5.0, Palisade Corporation,
             New York, NY.
       Ramanathan, R. (2001), “A data envelopment analysis of comparative performance of schools in
             the Netherlands”, Operations Research, Vol. 38 No. 2, pp. 160-81.
       Ray, C.S. and Jeon, Y. (2003), “Reputation and efficiency: a nonparametric assessment of
             America’s top-rated MBA programs”, Working Paper 2003-13, March, available at: www.
             econ.uconn.edu/ (accessed 23 September 2009).
       Sahay, B.S. and Thakur, R. (2007), “Excellence through accreditation in Indian B-Schools”,
             Global Journal of Flexible Systems in Management, Vol. 8 No. 4, pp. 9-16.
       Sahay, B.S. and Thakur, R. (2008), “Making Indian management education globally competitive”,
             Proceedings of ASBBS, Vol. 15 No. 1, pp. 1332-9.
       Specht, D.F. (1991), “A general regression neural network”, IEEE Transactions on Neural
             Networks, Vol. 2 No. 6, pp. 568-76.
       Sreekumar, S. and Patel, G. (2007), “Comparative analysis of B-school rankings and an alternate
             ranking method”, International Journal of Operations and Quantitative Management,
             Vol. 13 No. 1, pp. 33-46.
       Tam, K.Y. and Kiang, M.Y. (1992), “Managerial applications of neural networks: the case of bank
             failure predictions”, Management Science, Vol. 38 No. 7, pp. 926-47.
       Tomkins, C.Y. and Green, R. (1988), “An experiment in the use of data envelopment analysis of
             evaluating the efficiency of UK university departments of accounting”, Financial
             Accountability & Management, Vol. 14 No. 2, pp. 147-64.
       Wadhwa, S., Kumar, A. and Saxena, A. (2005), “Modeling and analysis of technical education
             system: a KM and DEA based approach”, Studies in Informatics and Control, Vol. 14 No. 4,
             pp. 235-50.
       Wang, Z. (1994), “An artificial neural network model for comparative study of education system
             of China”, Control Engineering Practice, Vol. 2 No. 1, pp. 167-80.
Whitley, R., Thomas, A. and Marceau, J. (1981), Masters of Business? Business Schools and              Indian business
      Business Graduates in Britain and France, Tavistock, London.
Wu, P., Fang, S-C., King, R.E. and Nuttle, H.L. (1995), “Decision surface modeling of apparel retail
                                                                                                               schools
      operations using neural network technology”, International Journal of Operations and
      Quantitative Management, Vol. 1 No. 1, pp. 33-47.

About the authors                                                                                                239
S. Sreekumar is an Associate Professor in Rourkela Institute of Management Studies, Rourkela
769015, India. His areas of interest include application of DEA for efficiency analysis and
multi-criteria decision making. He has 17 years of teaching experience in the areas of quantitative
techniques and information science. He has published 30 papers in various international and
national journals and conferences. He has also authored two books.
    S.S. Mahapatra is Professor in the Department of Mechanical Engineering, National Institute of
Technology Rourkela, India. He has more than 20 years of experience in teaching and research.
His current area of research includes multi-criteria decision making, quality engineering, assembly
line balancing, group technology, neural networks, and non-traditional optimization and simulation.
He has published more than 40 papers in referred journals. He has written few books related to his
research work. He is also currently dealing with few sponsored projects. S.S. Mahapatra is the
corresponding author and can be contacted at: mahapatrass2003@yahoo.com




To purchase reprints of this article please e-mail: reprints@emeraldinsight.com
Or visit our web site for further details: www.emeraldinsight.com/reprints

Contenu connexe

Tendances

EMPLOYEE ATTRITION PREDICTION IN INDUSTRY USING MACHINE LEARNING TECHNIQUES
EMPLOYEE ATTRITION PREDICTION IN INDUSTRY USING MACHINE LEARNING TECHNIQUESEMPLOYEE ATTRITION PREDICTION IN INDUSTRY USING MACHINE LEARNING TECHNIQUES
EMPLOYEE ATTRITION PREDICTION IN INDUSTRY USING MACHINE LEARNING TECHNIQUESIAEME Publication
 
Maintenance management practices_for_building
Maintenance management practices_for_buildingMaintenance management practices_for_building
Maintenance management practices_for_buildingkhairilhakimi12
 
Juxt Consult India Employee Speak 2007 Current Hr Trends
Juxt Consult India Employee Speak 2007 Current Hr TrendsJuxt Consult India Employee Speak 2007 Current Hr Trends
Juxt Consult India Employee Speak 2007 Current Hr TrendsJuxtConsult
 
The Effect of Information Technology, User Technical Skills, Education and Tr...
The Effect of Information Technology, User Technical Skills, Education and Tr...The Effect of Information Technology, User Technical Skills, Education and Tr...
The Effect of Information Technology, User Technical Skills, Education and Tr...AJHSSR Journal
 
Klibel5 bus 42
Klibel5 bus 42Klibel5 bus 42
Klibel5 bus 42KLIBEL
 
D472938.pdf
D472938.pdfD472938.pdf
D472938.pdfaijbm
 
The causality relationship between management in supply chain collaboration w...
The causality relationship between management in supply chain collaboration w...The causality relationship between management in supply chain collaboration w...
The causality relationship between management in supply chain collaboration w...Alexander Decker
 
Klibel5 econ 22_
Klibel5 econ 22_Klibel5 econ 22_
Klibel5 econ 22_KLIBEL
 
J478689.pdf
J478689.pdfJ478689.pdf
J478689.pdfaijbm
 
Klibel5 acc 36_
Klibel5 acc 36_Klibel5 acc 36_
Klibel5 acc 36_KLIBEL
 
Klibel5 bus 40
Klibel5 bus 40Klibel5 bus 40
Klibel5 bus 40KLIBEL
 
Klibel5 bus 19
Klibel5 bus 19Klibel5 bus 19
Klibel5 bus 19KLIBEL
 
AN ANALYSIS OF INCOME AND EXPENDITURE WITH SPECIAL REFERENCE TO BMTC, BANGALORE
AN ANALYSIS OF INCOME AND EXPENDITURE WITH SPECIAL REFERENCE TO BMTC, BANGALOREAN ANALYSIS OF INCOME AND EXPENDITURE WITH SPECIAL REFERENCE TO BMTC, BANGALORE
AN ANALYSIS OF INCOME AND EXPENDITURE WITH SPECIAL REFERENCE TO BMTC, BANGALOREIAEME Publication
 
Klibel5 acc 39_
Klibel5 acc 39_Klibel5 acc 39_
Klibel5 acc 39_KLIBEL
 
Klibel5 acc 38_
Klibel5 acc 38_Klibel5 acc 38_
Klibel5 acc 38_KLIBEL
 
A Study on Identification of the Employability Skills Level among Arts and Sc...
A Study on Identification of the Employability Skills Level among Arts and Sc...A Study on Identification of the Employability Skills Level among Arts and Sc...
A Study on Identification of the Employability Skills Level among Arts and Sc...inventionjournals
 
Klibel5 bus 15
Klibel5 bus 15Klibel5 bus 15
Klibel5 bus 15KLIBEL
 

Tendances (20)

EMPLOYEE ATTRITION PREDICTION IN INDUSTRY USING MACHINE LEARNING TECHNIQUES
EMPLOYEE ATTRITION PREDICTION IN INDUSTRY USING MACHINE LEARNING TECHNIQUESEMPLOYEE ATTRITION PREDICTION IN INDUSTRY USING MACHINE LEARNING TECHNIQUES
EMPLOYEE ATTRITION PREDICTION IN INDUSTRY USING MACHINE LEARNING TECHNIQUES
 
Icplt proceedings
Icplt proceedingsIcplt proceedings
Icplt proceedings
 
The Effect of Non-Statutory Welfare Schemes on the Motivation Levels of Non-A...
The Effect of Non-Statutory Welfare Schemes on the Motivation Levels of Non-A...The Effect of Non-Statutory Welfare Schemes on the Motivation Levels of Non-A...
The Effect of Non-Statutory Welfare Schemes on the Motivation Levels of Non-A...
 
Maintenance management practices_for_building
Maintenance management practices_for_buildingMaintenance management practices_for_building
Maintenance management practices_for_building
 
Juxt Consult India Employee Speak 2007 Current Hr Trends
Juxt Consult India Employee Speak 2007 Current Hr TrendsJuxt Consult India Employee Speak 2007 Current Hr Trends
Juxt Consult India Employee Speak 2007 Current Hr Trends
 
The Effect of Information Technology, User Technical Skills, Education and Tr...
The Effect of Information Technology, User Technical Skills, Education and Tr...The Effect of Information Technology, User Technical Skills, Education and Tr...
The Effect of Information Technology, User Technical Skills, Education and Tr...
 
Klibel5 bus 42
Klibel5 bus 42Klibel5 bus 42
Klibel5 bus 42
 
D472938.pdf
D472938.pdfD472938.pdf
D472938.pdf
 
The causality relationship between management in supply chain collaboration w...
The causality relationship between management in supply chain collaboration w...The causality relationship between management in supply chain collaboration w...
The causality relationship between management in supply chain collaboration w...
 
Klibel5 econ 22_
Klibel5 econ 22_Klibel5 econ 22_
Klibel5 econ 22_
 
J478689.pdf
J478689.pdfJ478689.pdf
J478689.pdf
 
Ijebea14 240
Ijebea14 240Ijebea14 240
Ijebea14 240
 
Klibel5 acc 36_
Klibel5 acc 36_Klibel5 acc 36_
Klibel5 acc 36_
 
Klibel5 bus 40
Klibel5 bus 40Klibel5 bus 40
Klibel5 bus 40
 
Klibel5 bus 19
Klibel5 bus 19Klibel5 bus 19
Klibel5 bus 19
 
AN ANALYSIS OF INCOME AND EXPENDITURE WITH SPECIAL REFERENCE TO BMTC, BANGALORE
AN ANALYSIS OF INCOME AND EXPENDITURE WITH SPECIAL REFERENCE TO BMTC, BANGALOREAN ANALYSIS OF INCOME AND EXPENDITURE WITH SPECIAL REFERENCE TO BMTC, BANGALORE
AN ANALYSIS OF INCOME AND EXPENDITURE WITH SPECIAL REFERENCE TO BMTC, BANGALORE
 
Klibel5 acc 39_
Klibel5 acc 39_Klibel5 acc 39_
Klibel5 acc 39_
 
Klibel5 acc 38_
Klibel5 acc 38_Klibel5 acc 38_
Klibel5 acc 38_
 
A Study on Identification of the Employability Skills Level among Arts and Sc...
A Study on Identification of the Employability Skills Level among Arts and Sc...A Study on Identification of the Employability Skills Level among Arts and Sc...
A Study on Identification of the Employability Skills Level among Arts and Sc...
 
Klibel5 bus 15
Klibel5 bus 15Klibel5 bus 15
Klibel5 bus 15
 

En vedette

4.the singularity
4.the singularity4.the singularity
4.the singularitylibfsb
 
Chapter 01 power_point
Chapter 01 power_pointChapter 01 power_point
Chapter 01 power_pointncash513
 
6.benchmarking of
6.benchmarking of6.benchmarking of
6.benchmarking oflibfsb
 
Chị hằng viện ta tiếng đồn gần xa đẹp như trăng rằm
Chị hằng viện ta tiếng đồn gần xa đẹp như trăng rằmChị hằng viện ta tiếng đồn gần xa đẹp như trăng rằm
Chị hằng viện ta tiếng đồn gần xa đẹp như trăng rằmlibfsb
 
Going Out For Dinner
Going Out For DinnerGoing Out For Dinner
Going Out For Dinnermfern437
 
Principles of food beverage and labor cost controls
Principles of food  beverage  and labor cost controlsPrinciples of food  beverage  and labor cost controls
Principles of food beverage and labor cost controlslibfsb
 
Menu management week 4 31 oct 2012
Menu management week 4  31 oct 2012Menu management week 4  31 oct 2012
Menu management week 4 31 oct 2012Dimitris Dimitriou
 
The bar & beverage book
The bar & beverage bookThe bar & beverage book
The bar & beverage booklibfsb
 
1.the recession,
1.the recession,1.the recession,
1.the recession,libfsb
 
Foodbeverage
FoodbeverageFoodbeverage
Foodbeveragelibfsb
 
Introduction.to.management.in.the.hospitality.industry.10th.edition
Introduction.to.management.in.the.hospitality.industry.10th.editionIntroduction.to.management.in.the.hospitality.industry.10th.edition
Introduction.to.management.in.the.hospitality.industry.10th.editionlibfsb
 
Business Basics Teacher's Book
Business Basics Teacher's BookBusiness Basics Teacher's Book
Business Basics Teacher's Bookpepitagimenez7
 
Food and beverage_operations
Food and beverage_operationsFood and beverage_operations
Food and beverage_operationslibfsb
 
At the restaurant
At the restaurantAt the restaurant
At the restaurantmartamr86
 

En vedette (15)

4.the singularity
4.the singularity4.the singularity
4.the singularity
 
CA1, CH4, Notes
CA1, CH4, NotesCA1, CH4, Notes
CA1, CH4, Notes
 
Chapter 01 power_point
Chapter 01 power_pointChapter 01 power_point
Chapter 01 power_point
 
6.benchmarking of
6.benchmarking of6.benchmarking of
6.benchmarking of
 
Chị hằng viện ta tiếng đồn gần xa đẹp như trăng rằm
Chị hằng viện ta tiếng đồn gần xa đẹp như trăng rằmChị hằng viện ta tiếng đồn gần xa đẹp như trăng rằm
Chị hằng viện ta tiếng đồn gần xa đẹp như trăng rằm
 
Going Out For Dinner
Going Out For DinnerGoing Out For Dinner
Going Out For Dinner
 
Principles of food beverage and labor cost controls
Principles of food  beverage  and labor cost controlsPrinciples of food  beverage  and labor cost controls
Principles of food beverage and labor cost controls
 
Menu management week 4 31 oct 2012
Menu management week 4  31 oct 2012Menu management week 4  31 oct 2012
Menu management week 4 31 oct 2012
 
The bar & beverage book
The bar & beverage bookThe bar & beverage book
The bar & beverage book
 
1.the recession,
1.the recession,1.the recession,
1.the recession,
 
Foodbeverage
FoodbeverageFoodbeverage
Foodbeverage
 
Introduction.to.management.in.the.hospitality.industry.10th.edition
Introduction.to.management.in.the.hospitality.industry.10th.editionIntroduction.to.management.in.the.hospitality.industry.10th.edition
Introduction.to.management.in.the.hospitality.industry.10th.edition
 
Business Basics Teacher's Book
Business Basics Teacher's BookBusiness Basics Teacher's Book
Business Basics Teacher's Book
 
Food and beverage_operations
Food and beverage_operationsFood and beverage_operations
Food and beverage_operations
 
At the restaurant
At the restaurantAt the restaurant
At the restaurant
 

Similaire à 3.performance modeling

Measuring Technical and Scale Efficiency of Banks in India Using DEA
Measuring Technical and Scale Efficiency of Banks in India Using DEAMeasuring Technical and Scale Efficiency of Banks in India Using DEA
Measuring Technical and Scale Efficiency of Banks in India Using DEAiosrjce
 
Employee perception towards effective training program a study on some select...
Employee perception towards effective training program a study on some select...Employee perception towards effective training program a study on some select...
Employee perception towards effective training program a study on some select...Alexander Decker
 
Entrepreneurial Competencies as a Tool for Improve the Enterprise Growth and ...
Entrepreneurial Competencies as a Tool for Improve the Enterprise Growth and ...Entrepreneurial Competencies as a Tool for Improve the Enterprise Growth and ...
Entrepreneurial Competencies as a Tool for Improve the Enterprise Growth and ...ijtsrd
 
Making Management Education in India Meet Global Quality
Making Management Education in India Meet Global QualityMaking Management Education in India Meet Global Quality
Making Management Education in India Meet Global Qualitypaperpublications3
 
IRJET- A Study on Emerging Trends in Performance Appraisal IN 21st ERA
IRJET-  	  A Study on Emerging Trends in Performance Appraisal IN 21st ERAIRJET-  	  A Study on Emerging Trends in Performance Appraisal IN 21st ERA
IRJET- A Study on Emerging Trends in Performance Appraisal IN 21st ERAIRJET Journal
 
Study of Performance Appraisal System for Faculty Members in Selected Managem...
Study of Performance Appraisal System for Faculty Members in Selected Managem...Study of Performance Appraisal System for Faculty Members in Selected Managem...
Study of Performance Appraisal System for Faculty Members in Selected Managem...ijtsrd
 
mm bagali....... skills phD..... Skills..... Competency .........Skills synop...
mm bagali....... skills phD..... Skills..... Competency .........Skills synop...mm bagali....... skills phD..... Skills..... Competency .........Skills synop...
mm bagali....... skills phD..... Skills..... Competency .........Skills synop...dr m m bagali, phd in hr
 
052 om c-dhanapal&gganesan-measuring_operational_efficiency_of (1) (1)
052 om c-dhanapal&gganesan-measuring_operational_efficiency_of (1) (1)052 om c-dhanapal&gganesan-measuring_operational_efficiency_of (1) (1)
052 om c-dhanapal&gganesan-measuring_operational_efficiency_of (1) (1)Anil Aks
 
Making Management Education in India Meet Global Quality
Making Management Education in India Meet Global QualityMaking Management Education in India Meet Global Quality
Making Management Education in India Meet Global Qualitypaperpublications3
 
An Analysis on Employee Turnover Problem in Construction Industry
An Analysis on Employee Turnover Problem in Construction IndustryAn Analysis on Employee Turnover Problem in Construction Industry
An Analysis on Employee Turnover Problem in Construction Industryijtsrd
 
A Mediated Model of Employee commitment: The Impact of Knowledge Management P...
A Mediated Model of Employee commitment: The Impact of Knowledge Management P...A Mediated Model of Employee commitment: The Impact of Knowledge Management P...
A Mediated Model of Employee commitment: The Impact of Knowledge Management P...IJAEMSJORNAL
 
Dynamics of Motivation strategies for Knowledgeworker
Dynamics of Motivation strategies for KnowledgeworkerDynamics of Motivation strategies for Knowledgeworker
Dynamics of Motivation strategies for KnowledgeworkerIJERA Editor
 
The Impact of Intellectual Capital on Firm Performance of Manufacturing SMEs ...
The Impact of Intellectual Capital on Firm Performance of Manufacturing SMEs ...The Impact of Intellectual Capital on Firm Performance of Manufacturing SMEs ...
The Impact of Intellectual Capital on Firm Performance of Manufacturing SMEs ...IIJSRJournal
 
Study the Role of Strategic Analysis of Business Schools – North Zone.
Study the Role of Strategic Analysis of Business Schools – North Zone.Study the Role of Strategic Analysis of Business Schools – North Zone.
Study the Role of Strategic Analysis of Business Schools – North Zone.Muhammed Anaz PK
 
BUSINESS SCHOOL MAKEOVER; A INDUSTRY PERSPECTIVE
BUSINESS SCHOOL MAKEOVER; A INDUSTRY PERSPECTIVEBUSINESS SCHOOL MAKEOVER; A INDUSTRY PERSPECTIVE
BUSINESS SCHOOL MAKEOVER; A INDUSTRY PERSPECTIVEIJITE
 
BUSINESS SCHOOL MAKEOVER; A INDUSTRY PERSPECTIVE
BUSINESS SCHOOL MAKEOVER; A INDUSTRY PERSPECTIVEBUSINESS SCHOOL MAKEOVER; A INDUSTRY PERSPECTIVE
BUSINESS SCHOOL MAKEOVER; A INDUSTRY PERSPECTIVEIJITE
 
A REVIEW ON PERFORMANCE MANAGEMENT AND APPRAISAL IN CONSTRUCTION INDUSTRY TOW...
A REVIEW ON PERFORMANCE MANAGEMENT AND APPRAISAL IN CONSTRUCTION INDUSTRY TOW...A REVIEW ON PERFORMANCE MANAGEMENT AND APPRAISAL IN CONSTRUCTION INDUSTRY TOW...
A REVIEW ON PERFORMANCE MANAGEMENT AND APPRAISAL IN CONSTRUCTION INDUSTRY TOW...Dereck Downing
 

Similaire à 3.performance modeling (20)

Trn 09
Trn 09Trn 09
Trn 09
 
Measuring Technical and Scale Efficiency of Banks in India Using DEA
Measuring Technical and Scale Efficiency of Banks in India Using DEAMeasuring Technical and Scale Efficiency of Banks in India Using DEA
Measuring Technical and Scale Efficiency of Banks in India Using DEA
 
CHAPTER 4.pdf
CHAPTER 4.pdfCHAPTER 4.pdf
CHAPTER 4.pdf
 
Employee perception towards effective training program a study on some select...
Employee perception towards effective training program a study on some select...Employee perception towards effective training program a study on some select...
Employee perception towards effective training program a study on some select...
 
H045074150
H045074150H045074150
H045074150
 
Entrepreneurial Competencies as a Tool for Improve the Enterprise Growth and ...
Entrepreneurial Competencies as a Tool for Improve the Enterprise Growth and ...Entrepreneurial Competencies as a Tool for Improve the Enterprise Growth and ...
Entrepreneurial Competencies as a Tool for Improve the Enterprise Growth and ...
 
Making Management Education in India Meet Global Quality
Making Management Education in India Meet Global QualityMaking Management Education in India Meet Global Quality
Making Management Education in India Meet Global Quality
 
IRJET- A Study on Emerging Trends in Performance Appraisal IN 21st ERA
IRJET-  	  A Study on Emerging Trends in Performance Appraisal IN 21st ERAIRJET-  	  A Study on Emerging Trends in Performance Appraisal IN 21st ERA
IRJET- A Study on Emerging Trends in Performance Appraisal IN 21st ERA
 
Study of Performance Appraisal System for Faculty Members in Selected Managem...
Study of Performance Appraisal System for Faculty Members in Selected Managem...Study of Performance Appraisal System for Faculty Members in Selected Managem...
Study of Performance Appraisal System for Faculty Members in Selected Managem...
 
mm bagali....... skills phD..... Skills..... Competency .........Skills synop...
mm bagali....... skills phD..... Skills..... Competency .........Skills synop...mm bagali....... skills phD..... Skills..... Competency .........Skills synop...
mm bagali....... skills phD..... Skills..... Competency .........Skills synop...
 
052 om c-dhanapal&gganesan-measuring_operational_efficiency_of (1) (1)
052 om c-dhanapal&gganesan-measuring_operational_efficiency_of (1) (1)052 om c-dhanapal&gganesan-measuring_operational_efficiency_of (1) (1)
052 om c-dhanapal&gganesan-measuring_operational_efficiency_of (1) (1)
 
Making Management Education in India Meet Global Quality
Making Management Education in India Meet Global QualityMaking Management Education in India Meet Global Quality
Making Management Education in India Meet Global Quality
 
An Analysis on Employee Turnover Problem in Construction Industry
An Analysis on Employee Turnover Problem in Construction IndustryAn Analysis on Employee Turnover Problem in Construction Industry
An Analysis on Employee Turnover Problem in Construction Industry
 
A Mediated Model of Employee commitment: The Impact of Knowledge Management P...
A Mediated Model of Employee commitment: The Impact of Knowledge Management P...A Mediated Model of Employee commitment: The Impact of Knowledge Management P...
A Mediated Model of Employee commitment: The Impact of Knowledge Management P...
 
Dynamics of Motivation strategies for Knowledgeworker
Dynamics of Motivation strategies for KnowledgeworkerDynamics of Motivation strategies for Knowledgeworker
Dynamics of Motivation strategies for Knowledgeworker
 
The Impact of Intellectual Capital on Firm Performance of Manufacturing SMEs ...
The Impact of Intellectual Capital on Firm Performance of Manufacturing SMEs ...The Impact of Intellectual Capital on Firm Performance of Manufacturing SMEs ...
The Impact of Intellectual Capital on Firm Performance of Manufacturing SMEs ...
 
Study the Role of Strategic Analysis of Business Schools – North Zone.
Study the Role of Strategic Analysis of Business Schools – North Zone.Study the Role of Strategic Analysis of Business Schools – North Zone.
Study the Role of Strategic Analysis of Business Schools – North Zone.
 
BUSINESS SCHOOL MAKEOVER; A INDUSTRY PERSPECTIVE
BUSINESS SCHOOL MAKEOVER; A INDUSTRY PERSPECTIVEBUSINESS SCHOOL MAKEOVER; A INDUSTRY PERSPECTIVE
BUSINESS SCHOOL MAKEOVER; A INDUSTRY PERSPECTIVE
 
BUSINESS SCHOOL MAKEOVER; A INDUSTRY PERSPECTIVE
BUSINESS SCHOOL MAKEOVER; A INDUSTRY PERSPECTIVEBUSINESS SCHOOL MAKEOVER; A INDUSTRY PERSPECTIVE
BUSINESS SCHOOL MAKEOVER; A INDUSTRY PERSPECTIVE
 
A REVIEW ON PERFORMANCE MANAGEMENT AND APPRAISAL IN CONSTRUCTION INDUSTRY TOW...
A REVIEW ON PERFORMANCE MANAGEMENT AND APPRAISAL IN CONSTRUCTION INDUSTRY TOW...A REVIEW ON PERFORMANCE MANAGEMENT AND APPRAISAL IN CONSTRUCTION INDUSTRY TOW...
A REVIEW ON PERFORMANCE MANAGEMENT AND APPRAISAL IN CONSTRUCTION INDUSTRY TOW...
 

Plus de libfsb

Principles of food beverage and labor cost controls
Principles of food  beverage  and labor cost controlsPrinciples of food  beverage  and labor cost controls
Principles of food beverage and labor cost controlslibfsb
 
Food safety basics a reference guide for foodservice operators
Food safety basics a reference guide for foodservice operatorsFood safety basics a reference guide for foodservice operators
Food safety basics a reference guide for foodservice operatorslibfsb
 
The bar & beverage book
The bar & beverage bookThe bar & beverage book
The bar & beverage booklibfsb
 
Hotel front office management 3rd edition
Hotel front office management 3rd editionHotel front office management 3rd edition
Hotel front office management 3rd editionlibfsb
 
3.great profits
3.great profits3.great profits
3.great profitslibfsb
 
2.pleasing all
2.pleasing all2.pleasing all
2.pleasing alllibfsb
 
9.greener library
9.greener library9.greener library
9.greener librarylibfsb
 
8.moving on
8.moving on 8.moving on
8.moving on libfsb
 
7.let them
7.let them7.let them
7.let themlibfsb
 
6.dealing with
6.dealing with6.dealing with
6.dealing withlibfsb
 
5.the management
5.the management5.the management
5.the managementlibfsb
 
4.making the
4.making the4.making the
4.making thelibfsb
 
2.free electronic
2.free electronic2.free electronic
2.free electroniclibfsb
 
13.roi. measuring
13.roi. measuring13.roi. measuring
13.roi. measuringlibfsb
 
11.the yogi
11.the yogi11.the yogi
11.the yogilibfsb
 
10.efficiencies and
10.efficiencies and10.efficiencies and
10.efficiencies andlibfsb
 
9.the value
9.the value9.the value
9.the valuelibfsb
 
8.e books- little
8.e books- little8.e books- little
8.e books- littlelibfsb
 
7.a more
7.a more7.a more
7.a morelibfsb
 
Ie mba#16
Ie mba#16Ie mba#16
Ie mba#16libfsb
 

Plus de libfsb (20)

Principles of food beverage and labor cost controls
Principles of food  beverage  and labor cost controlsPrinciples of food  beverage  and labor cost controls
Principles of food beverage and labor cost controls
 
Food safety basics a reference guide for foodservice operators
Food safety basics a reference guide for foodservice operatorsFood safety basics a reference guide for foodservice operators
Food safety basics a reference guide for foodservice operators
 
The bar & beverage book
The bar & beverage bookThe bar & beverage book
The bar & beverage book
 
Hotel front office management 3rd edition
Hotel front office management 3rd editionHotel front office management 3rd edition
Hotel front office management 3rd edition
 
3.great profits
3.great profits3.great profits
3.great profits
 
2.pleasing all
2.pleasing all2.pleasing all
2.pleasing all
 
9.greener library
9.greener library9.greener library
9.greener library
 
8.moving on
8.moving on 8.moving on
8.moving on
 
7.let them
7.let them7.let them
7.let them
 
6.dealing with
6.dealing with6.dealing with
6.dealing with
 
5.the management
5.the management5.the management
5.the management
 
4.making the
4.making the4.making the
4.making the
 
2.free electronic
2.free electronic2.free electronic
2.free electronic
 
13.roi. measuring
13.roi. measuring13.roi. measuring
13.roi. measuring
 
11.the yogi
11.the yogi11.the yogi
11.the yogi
 
10.efficiencies and
10.efficiencies and10.efficiencies and
10.efficiencies and
 
9.the value
9.the value9.the value
9.the value
 
8.e books- little
8.e books- little8.e books- little
8.e books- little
 
7.a more
7.a more7.a more
7.a more
 
Ie mba#16
Ie mba#16Ie mba#16
Ie mba#16
 

3.performance modeling

  • 1. The current issue and full text archive of this journal is available at www.emeraldinsight.com/1463-5771.htm Indian business Performance modeling of Indian schools business schools: a DEA-neural network approach 221 S. Sreekumar Rourkela Institute of Management Studies, Rourkela, India, and S.S. Mahapatra Department of Mechanical Engineering, National Institute of Technology, Rourkela, India Abstract Purpose – The main purpose of the present study is to develop an integrated approach combining data envelopment analysis (DEA) and neural network (NN) for assessment and prediction of performance of Indian B-schools for effective decision making as error and biasness due to human intervention in decision making is appreciably reduced. Design/methodology/approach – DEA, being a robust mathematical tool, has been employed to evaluate the efficiency of B-schools. DEA, basically, takes into account the input and output components of a decision-making unit (DMU) to calculate technical efficiency (TE). TE is treated as an indicator for performance of DMUs and comparison has been made among them. A sensitivity analysis has been carried out to study robustness of the ranking of schools obtained through DEA. Finally, NN is used to predict the efficiency when changes in inputs are caused due to market dynamism so that effective strategies can be evolved by the managers with limited available data. Findings – A total of 49 Indian B-schools are chosen for benchmarking purpose. The average score of efficiency is 0.625 with a standard deviation of 0.175 when Charnes, Cooper and Rhodes (CCR) model is used. Similarly, when the Banker, Charnes and Cooper (BCC) model is used the average score is 0.888 with a standard deviation of 0.063. The rank order correlation coefficient between the efficiency ranking obtained through CCR and BCC model is 0.736 ( p ¼ 0.000) which is significant. The peer group and peer weights for the inefficient B-schools have been identified. This is useful for benchmarking for the inefficient DMUs. They can identify the parameters in which they lack and take necessary steps for improvement. The peer group for the inefficient B-schools indicates the efficient B-schools to which the inefficient B-schools are closer in its combination of inputs and outputs. The TE obtained through DEA is used as output variable along with input variables considered in DEA as input and output parameters in a generalized regression NN during training phase. It can be observed that root mean square error is 0.009344 and 0.02323 for CCR- and BCC-efficiency prediction, respectively, during training. Similarly, root mean square error is 0.08585 and 0.03279 for CCR- and BCC-efficiency prediction, respectively, during testing. Now, individual schools can generate scenario with the data within their control and test their own performance through NN model. Originality/value – This work proposes integration of DEA and NN to assist the managers to predict the performance of an individual DMU based on input consumed and generate various “what-if” scenarios. The study provides a simple but comprehensive methodology for improving performance of B-schools in India. Keywords Benchmarking, Data analysis, Decision making units, Neural nets Paper type Research paper Benchmarking: An International Journal Vol. 18 No. 2, 2011 1. Introduction pp. 221-239 India has liberalized the business education market in 1990s resulting in a rapid q Emerald Group Publishing Limited 1463-5771 growth of business schools offering programs at both undergraduate as well as DOI 10.1108/14635771111121685
  • 2. BIJ post-graduate levels. The development of B-schools is largely adopting the policy of 18,2 self-sustainability and maximum of them are self-funded operated by private promoters. Most of the recruiters in India consider qualification in management as an added advantage. It caused demand in management education leading to an intense competition among the B-schools in the country. The low investment for entry and flourishing market has engineered the growth of B-schools throughout the country. 222 Mintzberg (1973) has pointed that the management school gives students degrees but it hardly teach them how to manage. Therefore, such degrees can barely be considered as prerequisites for managing firms in professional manner. On the contrary, Whitley et al. (1981) have advocated that many employers perceive holders of business education degree obviously distinguishes from those who do not possess it. Generally, students feel that getting a management degree from a reputed school may act as a formal way to batter career planning. Indian B-schools play a major role for providing career opportunity for around 68 percent of the Indian population who are in the 22-27 years age group. The quality of education imparted in Indian B-schools is reasonably good enough and many firms in the globe prefer Indian management graduates as a result of globalization and liberalization of the market economy. As a business strategy, manpower of cross-cultural nature may have edge over competitors (Sahay and Thakur, 2007). Dayal (2002) emphasized on changing the structure of management education in India and suggested a strategy for institutional development for upgrading the quality of the academic program. The Indian B-schools should understand the emerging context of the economy, the industries, business and their needs and work out what they are delivering today and what they are expected to deliver tomorrow (Sahay and Thakur, 2008). In this context, it is important for each of the Indian B-schools to know where one stands and design programs and pedagogy which will meet the future business needs. The new B-schools can set well-established institutes as their peers and follow them to become competitive. Although there is an unprecedented growth of schools in recent times, assessment on performance and efficiency of them is found to a limited extent in the literature. Measuring efficiency levels of the B-schools is an important issue for prospective students, parents, employers, and program administrators. Ranking of them can provide useful guidelines to all the stake holders involved in management education (Ojha, 2005). Some magazines like Outlook, Business World, Indian Management, etc. publish the annual report on the ranks of the Indian B-schools. However, ranking mechanism and sample size is questionable in such efforts. In this paper, a non-parametric technique called data envelopment analysis (DEA) is adopted to rank Indian B-schools based on their efficiency score. The scores can suggest inefficient and low-performing schools in an effective manner. Though the concept of benchmarking is good for improving the performance of individual unit, the problem associated with it is lack of transparency in data sharing. Therefore, methodology which will allow the individual schools to generate scenario with the data within their control and perform at the desired level is highly desirable. To address this problem, a neural network (NN) model is developed and trained to predict the performance level of the individual B-schools. The proposed model, in this paper, integrates the DEA and NN models to predict the performance of Indian B-schools. The data necessary for this study are collected from standard weekly business magazines and journals. The data sources employ professional surveying agencies for data collection. Therefore, data are considered to be reliable although collected from secondary sources. For the purpose
  • 3. of the confidentiality of the schools and avoidance of conflicting interests, the identities Indian business of schools are not disclosed. However, some of the top Indian B-schools such as Indian schools Institute of Management, Xavier Labour Relations Institute, Management Development Institute, and Faculty of Management Studies (Delhi University) have been included in the dataset. 2. Literature review 223 In the recent years, several studies have been undertaken for analysis of efficiency in education sector using DEA methodology. Each study differs in its scope, meaning, and definition of decision-making units (DMUs). Tomkins and Green (1988) conducted DEA analysis to test the performance of 20 accounting departments in UK. Johnes and Johnes (1993) investigated the use of DEA in the assessment of performance of university departments of the UK over the period 1984-1988. McMullen (1997) has applied DEA to assess the relative desirability of Association to Advanced Collegiate Schools of Business-accredited MBA programs. The authors have incorporated several attributes of MBA programs into the model for finding out most desirable program in terms of these attributes. McMillan and Datta (1998) have assessed the relative efficiency of 45 Canadian universities using DEA. A subset of universities including universities from each of three categories such as comprehensive with medical school, comprehensive without medical school, and primarily undergraduate are regularly found efficient while some universities exhibit inefficiency. But, overall efficiency for most of the universities is relatively high. Ramanathan (2001) has compared the performance of selected schools in The Netherlands using DEA and found that the efficiencies of the schools are closely related with their performance. The authors have also observed that several non-discretionary input variables can influence the efficiency scores but some of them are not in direct control of management of the school. Lopes and Lanzer (2002) have addressed the issue of performance evaluation, productivity, and quality of academic departments at a university using a DEA model for cross-evaluation between departments considering the dimensions of teaching, research, and service quality. The authors have observed zero correlation between department teaching, research, and service and weak correlation between research productivity and quality. Ray and Jeon (2003) in their study employed a measure of Pareto-Koopmans global efficiency to evaluate the efficiency levels of MBA programs in Business Week’s top-rated list. They computed input and output-oriented radial and non-radial efficiency measures for comparison purpose. Among three-tier groups, the schools from a higher tier group on an average are more efficient than those from lower tiers although variations in efficiency levels do occur within the same tier. In India, comparatively less studies have been conducted for performance evaluation of B-schools using DEA. Wadhwa et al. (2005) has proposed integration of DEA and knowledge management methods to evaluate the efficiency of technical education system (TES) in India. The authors claim that the suggested approach can assist decision makers in selecting proper institutes to further strengthen the TES in an efficient and effective manner. A number of successful business applications of artificial neural networks (ANNs) have been discussed in the literature, particularly in financial services (Tam and Kiang, 1992), transportation services (Nordmann and Luxhoj, 2000), telecommunications (Mozer et al., 1999), etc. Lu et al. (1996) have compared the effectiveness of NNs and the multinomial logit model, and concluded that the ANNs perform better than logit regressions in franchising decision making. Wu et al. (1995) have applied NN approach for the decision
  • 4. BIJ surface modeling of apparel retail operations. Tam and Kiang (1992) have discussed a back 18,2 propagation NN application in predicting bankruptcy of financial institutions based on financial ratios. Dutta and Shekhar (1988) have applied NNs to a generalization problem of predicting the corporate bond ratings. Chiang et al. (1996) have discussed a back propagation NN approach to mutual fund net asset value forecasting. Hu et al. (2004) have found that ANN can perform better than logistic regression in the modeling of foreign 224 equities. Kimoto et al. (1990) have applied modular NNs to develop a buying and selling timing prediction system for stocks on the Tokyo Stock Exchange using a high-speed learning method called supplementary learning. Odom and Sharda (1990) have developed an NN model using back propagation for prediction of bankruptcy and compared results with discriminant analysis. It is claimed that ANN model performs better than discriminant analysis which is generally used for such type of problems. In education sector, some ANN models have been reported for prediction of academic performance of educational institutions considering qualitative as well as quantitative criteria (Hoefer and Gould, 2000; Kannan, 2005; Naik and Ragothaman, 2004; Wang, 1994). However, the application of NNs to model qualitative and intangible aspects of different services is not addressed adequately in the literature of education. It may be apposite to extend implementation of NNs to address more general and theoretical issues in service sector, such as education. 3. Methodology 3.1 Data envelopment analysis DEA, introduced by Charnes et al. (1978), computes efficiency score of each unit by comparing the efficiency score of each unit with that of its peers. Geometrically, a frontier can be constructed comprising of best performers. The units lying on the frontier are said to be efficient, and other units are treated as inefficient. Algebraically, the DEA model can be written as: Pn Xn Xm ur yrj0 max hj0 ¼ Pr¼1 m subject to ur yrj0 2 vi xij0 # 0 ur ; vi $ 0 ;r; i ð1Þ i¼1 vi xij0 r¼1 i¼1 where: hj0 ¼ relative efficiency of target DMU j0. r ¼ 1, 2,. . .n the number of outputs. i ¼ 1, 2,. . .m the number of inputs. j ¼ 1, 2,. . .s the number of DMUs. ur ¼ weight attached to the output r. vi ¼ weight attached to the input i. yrjo ¼ quantity of rth output produced by the DMU Jo. xijo ¼ quantity of ith input consumed by the DMU Jo. The DEA models may have any of the two orientations viz. input orientation and output orientation. Input orientation means how much inputs can be reduced while maintaining the same level of output. But output orientation of DEA is characterized by how much output can be increased while keeping the level of inputs constant. The latter orientation is more relevant for many service providers where the objective is to maximize the output maintaining the same level of inputs.
  • 5. Another variation to a DEA model is the returns to scale (RTS) assumption. Constant, Indian business decreasing, increasing, and variable RTS assumptions may be employed. Constant return to scale (CRS) implies that doubling inputs will exactly double outputs. Decreasing return schools to scale implies that doubling inputs will less-than-double outputs. Increasing return to scale implies that doubling inputs will more-than-double outputs. Thus, variable return to scale (VRS) allows for a combination of constant, increasing, and decreasing inputs and outputs. The DEA model shown in Equation (1) assumes a CRS. The drawback with 225 the CRS model is that it compares DMUs only based on overall efficiency assuming constant RTS. It ignores the fact that different DMUs could be operating at different scales. To overcome this drawback, Banker et al. (1984) developed a model which considers variable RTS and compares DMUs purely on the basis of TE. The model can be shown as below: min u Xn subject to li xji 2 uxjj0 # 0 ;j i¼1 X n ð2Þ li yrj 2 yjj0 $ 0 ;r i¼1 li ¼ 1 ;i where: u ¼ efficiency score. li ¼ dual variable. The difference between the CRS model (1) and the VRS model (2) is that the li is restricted to one. This has the effect of removing the constraint in the CRS model that DMUs must be scale efficient. Consequently, the VRS model allows variable RTS and measures only TE for each DMU. Thus, a DMU to be considered as CRS efficient, it must be both scale and technical efficient. For a DMU to be considered VRS efficient, it only needs to be technically efficient. 3.2 Neural network An ANN is an information processing paradigm that is inspired by the way biological nervous systems, such as the human brain and process information. It is composed of a large number of highly interconnected processing elements (neurons) working in conjunction to solve specific problems. ANNs, like people, learn by example. An ANN is configured for a specific application such as pattern recognition or data classification through a learning process. Learning in biological systems involves adjustments to the connections that exist between the neurons which is true for ANNs as well. An NN consists of a network of neurons. Each neuron is associated with an input vector, a weight vector corresponding to the input vector, a scalar bias, a transfer function, and an output vector as shown in Figure 1. An NN may consist of one or more neurons in each layer. In a network, the final layer is called the output layer and all previous layers are called hidden layers. In the hidden layers, the output of a layer becomes the input for the following layer. The transfer function of a neuron converts the input to the output of the neuron. Multi-layer NNs are quite powerful tools used in solving many different complex problems. Various types of NNs are available for different purposes. In this study, a multi-layer back propagation NN architecture is adopted.
  • 6. BIJ Teach/use 18,2 W1 X W2 X Inputs Weights 226 Neuron Output Wn X Figure 1. A typical neuron Teaching input A typical NN is shown in Figure 2. There are three layers – a layer of “input” units is connected to a layer of “hidden” units, which is connected to a layer of “output” units. The behavior of the output units depends on the activity of the hidden units and the weights between the hidden and output units. The architectures of ANN may be single layer or multi-layer. In the single-layer organization, all units are connected to one another. It constitutes the most general case and is of more potential computational power than hierarchically structured multi-layer organizations. As discussed, NNs are capable of learning complex relationships in data. The problems NNs are used for can be divided in two general groups: classification problems in which one tries to determine what type of category an unknown item falls into and numeric problems where one attempts to predict a specific numeric outcome (Palisade Corporation, 2008). There are many computer software packages available for building and analysing NNs. In this work, Neural Tools Version 5.0 by Palisade Corporation (2008) is used. This software automatically scales all input data. Scaling involves mapping each variable to a range with minimum and maximum values of 0 and 1. A non-linear scaling function known as “tanh” is used as activation function. This function tends to squeeze data Input Hidden Output Input 1 Input 2 Output Input 3 Figure 2. A simple feed forward NN Input 4 with three layers
  • 7. together at the low and high ends of the original data range (Mostafa, 2009). An NN Indian business configuration called generalized regression neural networks (GRNN) put forward schools by Specht (1991) is adopted to give the best possible predictions. The rationale for choosing the GRNN configuration lies in the fact that it is a good numerical predictor and user need not to make decisions about the structure of a net. These nets always have two hidden layers of neurons, with one neuron per training case in the first hidden layer, and the size of the second layer determined by some facts about training data. 227 GRN architecture. A generalized regression neural (GRN) net for two independent numeric variables is structured as shown in Figure 3 with the assumption that there are just three training cases (Palisade Corporation, 2008). The pattern layer contains one node for each training case. Presenting a training case to the net consists here of presenting two independent numeric values. Each neuron in the pattern layer computes its distance from the presented case. The values passed to the numerator and denominator nodes are functions of the distance and the dependent value. The two nodes in the summation layer sum its inputs, while the output node divides them to generate the prediction. The distance function computed in the pattern layer neurons uses “smoothing factors”; every input has its own “smoothing factor” value. With a single input, the greater the value of the smoothing factor, the more significant distant training cases become for the predicted value. With two inputs, the smoothing factor relates to the distance along one axis on a plane, and in general, with multiple inputs, to one dimension in multi-dimensional space. Training a GRN net consists of optimizing smoothing factors to minimize the error on the training set, and the conjugate gradient descent optimization method is used to accomplish that. The error measure used during training to evaluate different sets of smoothing factors is the mean squared error. However, when computing the squared error for a training case, that case is temporarily excluded from the pattern layer. This is because the excluded neuron would compute a zero distance, making other neurons insignificant in the computation of the prediction. 3.3 Data and data classification For solving the benchmarking problem, 49 top B-schools of India are considered using convenience sampling method. Data on 11 parameters as listed below are collected from various secondary sources. The secondary source reference includes popular Indian magazines like Outlook, Business World, Indian Management, and B-school directories which publish the annual report on the ranks of the Indian B-schools (Table I). Output Figure 3. A simple GRN Inputs Pattern Summation architecture later layer
  • 8. BIJ The data collected on the above parameters are classified into two categories based on 18,2 their nature for DEA and NN application. The criteria of selection of inputs and outputs are quite subjective; there is no specific rule for determining the procedure for selection of inputs and outputs (Ramanathan, 2001). The classification of input and output is done as follows (Sreekumar and Patel, 2007): Input: 228 X1: IC X2: IF X3: FEE Output: Y1: II Y2: PP Y3: IL Y4: RS Y5: SS Y6: FS Y7: ECA Y8: SAL S. no. Parameter Abbreviation Explanation 1 Intellectual capital IC Faculty/student ratio, teaching experience of faculty, corporate experience of faculty/students, PhD/students ratio, faculty with PhD (abroad), books, research papers, and cases 2 Industry interface II Revenue from consultancy, revenue from management development programs, seminars, and workshops 3 Infrastructure and IF Area (in acres), built-up area, computers per batch, facilities amphitheatre class room, library books, electronic database, residential facilities, single occupancy room, and MDP hostel 4 International linkage IL Student exchange program and faculty exchange program 5 Placement PP Percentage of student placed, median salary, maximum performance salary, minimum salary, percentage of students placed abroad, and return on investment 6 Extra curricular ECA National-level events organized and awards won activities by students 7 Recruiters RS Application of knowledge of subject/skills, satisfaction analytical skills, communication and presentation skills, creativity, proactive attitude, and ability to work in team 8 Students satisfaction SS Satisfaction of ongoing students from the school 9 Faculty satisfaction FS Based on present faculty of the school Table I. 10 Fee FEE Fee collected from students List of parameters 11 Salary SAL Initial salary at which graduating students are placed
  • 9. In this study, the DEA and NN has been integrated to have an efficient predicting model. Indian business DEA, being a robust mathematical tool, has been employed to evaluate the efficiency of schools B-schools. DEA, basically, takes into account the input and output components of a DMU to calculate TE. TE is treated as an indicator for performance of DMUs and comparison has been made among them. A sensitivity analysis has been carried out to study robustness of the ranking of schools obtained through DEA. Finally, NN is used to predict the efficiency when changes in inputs are caused due to market dynamism so 229 that effective strategies can be evolved by the managers with limited available data. 4. Results and discussion The Charnes, Cooper and Rhodes (CCR)-DEA model as discussed above is based on constant RTS does not consider the size of B-school under consideration while calculating the efficiency. But in many cases the size of a unit may influence its ability to produce services more efficiently. So, we have also considered the VRS model for our study. The B-schools under consideration for our problem contain both private and government institute. The input for both the category of institute differs widely, so the output orientation model is used. It may be noted that the Banker, Charnes and Cooper (BCC) model allows variable RTS and measures only TE for each DMU whereas a DMU is considered as CCR efficient if it is both scale and technical efficient. The relative efficiency score of B-schools are analysed and presented in Table II. The BCC score is based on VRS assumption and measures the pure TE. The CCR score is based on CRS assumption and consist of non-additive combination of pure TE and scale efficiency. The table shows that in a scale of 0-1 the average score for the B-schools is 0.625 with a standard deviation of 0.175 when CCR model is used. Similarly, when the BCC model is used the average score is 0.888 with a standard deviation of 0.063. The rank-order correlation coefficient between the efficiency ranking obtained through CCR and BCC model is 0.736 ( p ¼ 0.000) which is significant. The above table also shows the peer group and peer weights for the inefficient B-schools. This is useful for benchmarking for the inefficient DMUs. They can identify the parameters in which they lack and take necessary steps for improvement. The peer group for the inefficient B-schools indicates the efficient B-schools to which the inefficient B-schools are closer in its combination of inputs and outputs. It may also be observed that in both CCR and BCC score there are multiple numbers of DMUs with efficiency score unity leading to tie case. The school which appears maximum number of times as peer in the above table may be treated as the best school. Moreover, this school is likely to be the school which is efficient with respect to a large number of factors, and is probably a good example of an exemplary operating performer. Efficient DMUs that appear seldom in the peer set of other inefficient DMUs are likely to possess a very uncommon input/output mix and are thus not suitable examples for other inefficient schools (Mostafa, 2009). Now, it is prudent to check the robustness of the model trough sensitivity. DEA is an extreme point technique because the efficiency frontier is formed by the actual performance of best-performing DMUs. A direct consequence of this aspect is that errors in measurement can affect the DEA result significantly. So, according to DEA technique, it is possible for a B-School to become efficient if it achieves exceptionally better results in terms of one output but performs below average in other outputs. The sensitivity of DEA efficiency can be verified by checking whether the efficiency of a DMU is affected appreciably:
  • 10. BIJ 18,2 230 oriented) Table II. Efficiency score (output DMU CCR Rank Peer Peer weights BCC Rank Peer Peer weights 1 1.000000 1 1 – 1.000000 1 1, 1 – 2 1.000000 1 2, 1 – 1.000000 1 2, 1 – 3 0.995190 5 1, 2, 6 0.634, 0.382, 2.71 £ 102 2 1.000000 1 3, 1 – 4 1.000000 1 4, 1 – 1.000000 1 4, 1 – 5 0.877809 7 1, 4 1.051, 0.306 1.000000 1 5, 1 – 6 1.000000 1 6, 1 – 1.000000 1 6, 1 – 7 0.799198 9 2, 4, 6 0.333, 0.674, 0.455 1.000000 1 7, 1 – 8 0.844716 8 2, 4, 6 0.341, 0.548, 0.296 0.964907 9 2, 4, 17 0.570, 0.330, 0.101 9 0.618562 20 1, 2, 6 0.159, 0.631, 0.741 0.907631 16 1, 2, 5, 6 0.111, 0.429, 0.177, 0.283 10 0.644462 19 4, 6 1.050, 0.491 0.898023 19 1, 2, 7 0.150, 0.508, 0.342 11 0.524947 31 2, 4, 6 0.704, 0.804, 0.333 0.902703 18 2, 5, 6 0.681, 0.289, 0.030 12 0.753229 10 2, 6 1.209, 1.11 £ 102 2 0.907649 15 2, 7 0.884, 0.115 13 0.584787 23 4, 6 1.141, 0.471 0.881374 21 1, 2, 7 0.487, 0.393, 0.120 14 0.729801 13 4, 6 0.659, 0.687 0.943011 11 2, 4, 6 0.486, 0.215, 0.299 15 0.514123 32 4, 6 1.147, 0.653 0.869822 25 2 1.000 16 0.526763 29 2, 4, 6 0.341, 1.090, 0.135 0.793427 47 1, 7 0.819, 0.181 17 0.904431 6 4, 6 1.00 £ 102 1, 1.079 1.000000 1 17 1.000 18 0.546956 26 4, 6 0.614, 1.227 0.905501 17 4, 6, 17 0.670, 0.226, 0.104 19 0.568486 24 4, 6 0.621, 1.141 0.921440 13 4, 6, 7, 17 0.366, 0.286, 0.301, 4.79 £ 102 2 20 0.508409 33 4, 6 0.353, 1.579 0.912566 14 6, 7 0.586, 0.414 21 0.556489 25 4, 6 0.634, 0.969 0.861150 28 2, 6 0.703, 0.297 22 0.504688 34 4, 6 0.905, 1.010 0.872109 24 2, 6, 7 0.692, 1.07 £ 102 2, 0.297 23 0.534655 27 4, 6 0.208, 1.415 0.856631 33 2, 6 0.266, 0.734 24 0.679945 15 4, 6 0.580, 0.744 0.857588 31 2, 6 0.624, 0.376 25 0.470277 44 4, 6 0.613, 1.356 0.843978 39 2, 4, 6 0.571, 0.136, 0.292 26 0.485306 39 4, 6 0.605, 1.349 0.869303 26 2, 4, 6, 7 0.533, 6.42 £ 102 2, 0.305, 9.74 £ 102 2 27 0.484885 40 4, 6 0.6135, 1.239 0.831249 41 2, 6 0.703, 0.297 28 0.472438 43 4, 6 1.044, 1.141 0.881026 22 2, 7 0.131, 0.869 29 0.500204 36 4, 6 0.701, 1.119 0.876344 23 2, 6 0.790, 0.210 30 0.487138 37 4, 6 0.567, 1.342 0.854651 34 2, 4, 6, 7 0.199, 0.238, 0.361, 0.202 31 0.476137 41 4, 6 0.353, 1.572 0.860350 30 2, 6, 7 0.110, 0.579, 0.311 32 0.475426 42 4, 6 0.479, 1.460 0.853220 37 2, 4, 6 0.453, 0.115, 0.431 (continued)
  • 11. DMU CCR Rank Peer Peer weights BCC Rank Peer Peer weights 33 0.486999 38 4, 6 0.444, 1.408 0.860563 29 2, 6 0.528, 0.472 34 0.525993 30 4, 6 0.744, 0.967 0.857388 32 2, 6 0.825, 0.175 35 0.665816 17 4, 6 9.41 £ 102 3, 1.453 0.945689 10 6, 7, 17 0.703, 1.38 £ 102 2, 0.283 36 0.613720 22 4, 6 0.313, 1.064 0.829688 42 2, 6 0.354, 0.6463 37 0.457457 45 4, 6 0.835, 1.010 0.806735 45 2, 6 0.930, 6.99 £ 102 2 38 0.533847 28 4, 6 0.272, 1.508 0.883930 20 2, 6 0.236, 0.358, 5.64 £ 102 2, 0.349 39 0.617969 21 4, 6 0.224, 1.199 0.854069 36 2, 4, 6 0.079, 0.181, 0.739 40 0.675052 16 4, 6 0.488, 0.814 0.854545 35 2, 6 0.528, 0.472 41 0.435924 47 4, 6 0.260, 1.778 0.838353 40 2, 4, 6 0.302, 5.05 £ 102 2, 0.648 42 0.452255 46 4, 6 0.403, 1.446 0.813703 44 2, 6 0.485, 0.515 43 0.739848 11 4, 6 9.72 £ 102 2, 1.221 0.938403 12 6, 7, 17 0.460, 6.65 £ 102 2, 0.473 44 0.500706 35 4, 6 0.373, 1.315 0.827064 43 2, 6 0.441, 0.559 45 0.695122 14 4 1.263 0.866262 27 4, 6, 17 8.72 £ 102 3, 0.882, 0.109 46 0.738727 12 4, 6 0.244, 0.926 0.848324 38 4, 6, 17 0.250, 0.677, 0.073 47 0.355991 49 4, 6 0.969, 1.364 0.792899 48 2 1.000 48 0.662169 18 4, 6 0.071636, 1.143831 0.800977 46 2, 6 9.17 £ 102 2, 0.908 49 0.412507 48 4, 6 0.776855, 1.205363 0.784348 49 2 0.882, 0.118 Notes: Avg (CCR) – 0.625297, Min – 0.355991, and SD – 0.175012; Avg (BCC) – 0.888339, Min – 0.784348, and SD – 0.062977; Pearson correlation of CCR and BCC ¼ 0.736, p-value ¼ 0.000 Indian business schools 231 Table II.
  • 12. BIJ . if only one input or output is omitted from DEA analysis; and 18,2 . dropping one efficient DMU at a time from DEA analysis. Initially the input “intellectual capital” is dropped from the analysis and TE of DMUs is calculated, then input “FEE” is dropped, and similarly the outputs “placement performance” is dropped from both CCR and BCC model. At the second level, the efficient unit DMU1 is 232 dropped to calculate the CCR and BCC efficiency. The results of both the stage are tabulated in Table III. The table shows that dropping the input “IC” and outputs “PP” one-by-one causes no significant change in the TE score of DMUs and efficient units are remaining efficient as such. Change in efficiency score is observed when the input “FEE” is dropped from the analysis. DMU6 is becoming inefficient when “FEE” is not considered for CCR efficiency. This indicates that “FEE” is an important input for such schools. At the second level of analysis, some of the efficient DMUs are dropped one-by-one. It is observed that the efficient units are remaining efficient as such when DMU1 is dropped from the DEA analysis but DMU3 becomes an efficient unit for CCR efficiency whereas there is no change efficiency status for BCC score. Finally, GRNN model is used to predict efficiency score of DMUs. The prediction results help the managers to use the available data for strategic decision making when the data for benchmarking is not shared by DMUs under consideration. The prediction can also generate various scenarios to guide the managers/administrators for effective decision making. In addition, human error is avoided during prediction process. Both CCR- and BCC-efficiency scores are predicted using GRNN model. The three inputs considered for prediction purpose are IC, IF, and FEE. As mentioned earlier, Neural Tools version 5.0 (Palisade Corporation, 2008) is used for building the network due to its flexibility and extensive capabilities. In general, 75 percent of the data is considered for training for mapping good relationship between inputs and outputs and 25 percent is used for testing. In this study, 39 cases are used for training the network chosen randomly from dataset to provide variation during training and ten cases for testing. The model is used to predict the output for the entire dataset of 49 B-schools. The network configuration setting is shown in Table IV. The network is run till a reasonable root mean square error is attained during training period. The number of epochs to obtain error within tolerance limit happens to be 69 and 98, respectively, for CCR- and BCC-efficiency prediction during training of the network. Once the network is well trained, it can be used for testing purpose as generalization capability of the network is ensured. It can be observed that root mean square error is 0.009344 and 0.02323 for CCR- and BCC-efficiency prediction, respectively, during training. Similarly, root mean square error is 0.08585 and 0.03279 for CCR- and BCC-efficiency prediction, respectively, during testing. The prediction results are shown in Table V. It can be observed that maximum absolute difference in CCR-efficiency score and the NN prediction is 0.108369 which occurs for DMU9. Similarly, maximum absolute difference in BCC-efficiency score and the NN prediction is 0.067631 which occurs again for DMU9. For rest of the cases, the absolute difference between either for CCR or BCC and neural network prediction is much smaller. The NN prediction of efficiency scores is compared with scores obtained through DEA using CCR and BCC in Figure 4. It is noted that the pattern is perfectly followed in both type of prediction.
  • 13. Efficiency Efficiency Dropping Dropping Dropping Dropping Dropping DMU Dropping DMU Dropping Dropping DMU CCR BCC IC, CCR IC, BCC FEE, CCR FEE, BCC 1, CCR 1, BCC PP, CCR PP, BCC 1 1.000000 1.000000 1.000000 1.000000 1.000000 1.000000 – – 1.000000 1.000000 2 1.000000 1.000000 1.000000 1.000000 1.000000 1.000000 1.000000 1.000000 1.000000 1.000000 3 0.995190 1.000000 0.981913 1.000000 0.970112 1.000000 1.000000 1.000000 0.995190 1.000000 4 1.000000 1.000000 1.000000 1.000000 1.000000 1.000000 1.000000 1.000000 1.000000 1.000000 5 0.877809 1.000000 0.772174 1.000000 0.877809 1.000000 0.981736 1.000000 0.877809 1.000000 6 1.000000 1.000000 1.000000 1.000000 0.586713 1.000000 1.000000 1.000000 1.000000 1.000000 7 0.799198 1.000000 0.774525 1.000000 0.607166 1.000000 0.799198 1.000000 0.799198 1.000000 8 0.844716 0.964907 0.844716 0.964907 0.673546 0.937684 0.844716 0.964907 0.844716 0.964907 9 0.618562 0.907631 0.618562 0.907631 0.392137 0.900923 0.620586 0.907877 0.618562 0.907631 10 0.644462 0.898023 0.644462 0.898023 0.433763 0.898023 0.644462 0.899259 0.644462 0.898023 11 0.524947 0.902703 0.492203 0.902703 0.430246 0.902703 0.524947 0.902703 0.524947 0.902703 12 0.753229 0.907649 0.522483 0.886576 0.738128 0.907649 0.753229 0.907649 0.753229 0.907649 13 0.584787 0.881374 0.584787 0.881374 0.436566 0.881374 0.584787 0.888147 0.584787 0.881374 14 0.729801 0.943011 0.729801 0.943011 0.423904 0.902508 0.729801 0.943011 0.729801 0.943011 15 0.514123 0.869822 0.514123 0.869822 0.376683 0.869822 0.514123 0.869822 0.514123 0.869822 16 0.526763 0.793427 0.526763 0.793427 0.434564 0.793427 0.526763 0.804730 0.526763 0.793427 17 0.904431 1.000000 0.904431 1.000000 0.355112 0.832466 0.904431 1.000000 0.904431 1.000000 18 0.546956 0.905501 0.546956 0.905501 0.262281 0.846514 0.546956 0.905501 0.546956 0.905501 19 0.568486 0.921440 0.568486 0.921440 0.320073 0.879471 0.568486 0.921440 0.568486 0.921440 20 0.508409 0.912566 0.508409 0.912566 0.274155 0.867030 0.508409 0.912566 0.508409 0.912566 21 0.556489 0.861150 0.556489 0.861150 0.331672 0.852071 0.556489 0.861150 0.556489 0.861150 22 0.504688 0.872109 0.504688 0.872109 0.324421 0.871695 0.504688 0.872109 0.504688 0.872109 23 0.534655 0.856631 0.534655 0.856631 0.265378 0.834320 0.534655 0.856631 0.534655 0.856631 24 0.679945 0.857588 0.679945 0.857588 0.359272 0.846154 0.679945 0.857588 0.679945 0.857588 25 0.470277 0.843978 0.470277 0.843978 0.249000 0.828402 0.470277 0.843978 0.470277 0.843978 26 0.485306 0.869303 0.485306 0.869303 0.270222 0.853810 0.485306 0.869303 0.485306 0.869303 27 0.484885 0.831249 0.484885 0.831249 0.257981 0.822485 0.484885 0.831249 0.484885 0.831249 28 0.472438 0.881026 0.472438 0.881026 0.311123 0.881026 0.472438 0.881026 0.472438 0.881026 29 0.500204 0.876344 0.500204 0.876344 0.273313 0.869822 0.500204 0.876344 0.500204 0.876344 30 0.487138 0.854651 0.487138 0.854651 0.260696 0.827372 0.487138 0.854651 0.487138 0.854651 31 0.476137 0.860350 0.476137 0.860350 0.249919 0.838258 0.476137 0.860350 0.476137 0.860350 (continued) report Sensitivity analysis Indian business schools Table III. 233
  • 14. BIJ 18,2 234 Table III. Efficiency Efficiency Dropping Dropping Dropping Dropping Dropping DMU Dropping DMU Dropping Dropping DMU CCR BCC IC, CCR IC, BCC FEE, CCR FEE, BCC 1, CCR 1, BCC PP, CCR PP, BCC 32 0.475426 0.853220 0.475426 0.853220 0.238986 0.834320 0.475426 0.853220 0.475426 0.853220 33 0.486999 0.860563 0.486999 0.860563 0.239228 0.846154 0.486999 0.860563 0.486999 0.860563 34 0.525993 0.857388 0.525993 0.857388 0.288255 0.852071 0.525993 0.857388 0.525993 0.857388 35 0.665816 0.945689 0.665816 0.945689 0.301387 0.849922 0.665816 0.945689 0.665816 0.945689 36 0.613720 0.829688 0.613720 0.829688 0.289507 0.810651 0.613720 0.829688 0.613720 0.829688 37 0.457457 0.806735 0.457457 0.806735 0.257054 0.804734 0.457457 0.806735 0.457457 0.806735 38 0.533847 0.883930 0.533847 0.883930 0.255412 0.804734 0.533847 0.883930 0.533847 0.883930 39 0.617969 0.854069 0.617969 0.854069 0.283359 0.822485 0.617969 0.854069 0.617969 0.854069 40 0.675052 0.854545 0.675052 0.854545 0.339557 0.840237 0.675052 0.854545 0.675052 0.854545 41 0.435924 0.838353 0.435924 0.838353 0.205649 0.816568 0.435924 0.838353 0.435924 0.838353 42 0.452255 0.813703 0.452255 0.813703 0.230183 0.798817 0.452255 0.813703 0.452255 0.813703 43 0.739848 0.938403 0.739848 0.938403 0.286491 0.835148 0.739848 0.938403 0.739848 0.938403 44 0.500706 0.827064 0.500706 0.827064 0.254029 0.810651 0.500706 0.827064 0.500706 0.827064 45 0.695122 0.866262 0.695122 0.866262 0.208424 0.822485 0.695122 0.866262 0.695122 0.866262 46 0.738727 0.848324 0.738727 0.848324 0.318954 0.804734 0.738727 0.848324 0.738727 0.848324 47 0.355991 0.792899 0.355991 0.792899 0.205402 0.792899 0.355991 0.792899 0.355991 0.792899 48 0.662169 0.800977 0.662169 0.800977 0.275852 0.775148 0.662169 0.800977 0.662169 0.800977 49 0.412507 0.784348 0.412507 0.784348 0.237532 0.781065 0.412507 0.784348 0.412507 0.784348
  • 15. Indian business CCR-efficiency prediction BCC-efficiency prediction schools Net information Configuration GRNN numeric predictor GRNN numeric predictor Independent category variables 0 0 Independent numeric variables 3 (IC, IF, and FEE) 3 (IC, IF, and FEE) Dependent variable Numeric variable (CCR-eff) Numeric variable (BCC-eff) 235 Training Number of cases 39 39 Number of trials 69 98 Bad predictions (%) (30 percent tolerance) 0.0000 0.0000 Root mean square error 0.009344 0.02323 Mean absolute error 0.005236 0.01606 Std deviation of abs. error 0.007740 0.01679 Testing Number of cases 10 10 Table IV. Bad predictions (%) (30 percent tolerance) 0.0000 0.0000 Network configuration Root mean square error 0.08585 0.03279 for CCR and BCC Mean absolute error 0.05931 0.02420 efficiency score Std deviation of absolute Error 0.06208 0.02212 prediction 5. Conclusions This study is concerned with the ranking of Indian B-schools using a non-parametric technique and developing an NN to predict the standing of schools based on limited number of parameters. Both CRS and VRS are considered to obtain the efficiency score of DMUs. The methodology facilitates in identifying the benchmarked institutions for the inefficient institutes. The process of benchmarking is useful in identifying the best business practices and formulating the winning strategies. The DEA methodology can be quite useful for Indian B-schools in identifying their position relative to their peers, and in formulating strategies for improvement by right mix of inputs and outputs. Although the concept of benchmarking is good for improving the performance of individual unit, the problem associated with it is lack of transparency in data sharing and data reliability. Some of the schools may not be too enthusiastic to share the data pertaining to their resource consumption and output produced. This requires a methodology which will allow the individual schools to generate scenario with the data within their control and test their own performance through simulation. For this purpose, an NN model is developed and trained to predict the performance level of the individual B-schools based on the input level consumed. The paper integrates the DEA model and NN. The output obtained through DEA model is used for training the NN for prediction of efficiency score of schools based on the input values. Like any other study, this paper also has several limitations giving opportunity for further research. The paper considers 11 parameters relevant for improving quality of Indian B-schools. The other pertinent factors like quality of inputs (students), investment pattern in the institution, funds generation by the institution, etc. could have been incorporated in the model for calculating efficiency score of schools. Next, some of the inputs may not be fully under the control of management leading to practically infeasible target. Again as the DEA gives the relative efficiency score, so it gets affected by sample size.
  • 16. BIJ NN Absolute NN Absolute 18,2 DMU CCR-eff prediction difference BCC-eff prediction difference 1 1.000000 0.995199 0.004801 1.000000 0.990000 0.010000 2 1.000000 0.999804 0.000196 1.000000 0.990000 0.010000 3 0.995190 0.995383 0.000190 1.000000 0.980000 0.020000 236 4 1.000000 1.000000 9 £ 102 8 1.000000 0.970000 0.030000 5 0.877809 0.877809 8.5 £ 102 10 1.000000 1.000000 0.000000 6 1.000000 1.000000 2.76 £ 102 7 1.000000 1.000000 0.000000 7 0.799198 0.798370 0.000828 1.000000 0.990000 0.010000 8 0.844716 0.844631 8.5 £ 102 5 0.964907 0.970000 0.005090 9 0.618562 0.510193 0.108369 0.907631 0.840000 0.067631 10 0.644462 0.643716 0.000746 0.898023 0.890000 0.008023 11 0.524947 0.524947 1.41 £ 102 7 0.902703 0.920000 0.017300 12 0.753229 0.753229 1.4 £ 102 12 0.907649 0.950000 0.042350 13 0.584787 0.585286 0.000500 0.881374 0.880000 0.001374 14 0.729801 0.711399 0.018402 0.943011 0.880000 0.063011 15 0.514123 0.515267 0.001140 0.869822 0.880000 0.010180 16 0.526763 0.638785 0.112020 0.793427 0.800000 0.006570 17 0.904431 0.904431 3.26 £ 102 7 1.000000 0.980000 0.020000 18 0.546956 0.515410 0.031546 0.905501 0.860000 0.045501 19 0.568486 0.489394 0.079092 0.921440 0.910000 0.011440 20 0.508409 0.501404 0.007005 0.912566 0.860000 0.052566 21 0.556489 0.556487 2.26 £ 102 6 0.861150 0.860000 0.001150 22 0.504688 0.504462 0.000226 0.872109 0.890000 0.017890 23 0.534655 0.536646 0.001990 0.856631 0.870000 0.013370 24 0.679945 0.698285 0.018340 0.857588 0.870000 0.012410 25 0.470277 0.478689 0.008410 0.843978 0.860000 0.016020 26 0.485306 0.482410 0.002896 0.869303 0.860000 0.009303 27 0.484885 0.491726 0.006840 0.831249 0.860000 0.028750 28 0.472438 0.472462 2.4 £ 102 5 0.881026 0.880000 0.001026 29 0.500204 0.501399 0.001190 0.876344 0.870000 0.006344 30 0.487138 0.486487 0.000651 0.854651 0.850000 0.004651 31 0.476137 0.484822 0.008680 0.860350 0.840000 0.020350 32 0.475426 0.484047 0.008620 0.853220 0.850000 0.003220 33 0.486999 0.491297 0.004300 0.860563 0.860000 0.000563 34 0.525993 0.511250 0.014743 0.857388 0.860000 0.002610 35 0.665816 0.662770 0.003046 0.945689 0.950000 0.004310 36 0.613720 0.616240 0.002520 0.829688 0.840000 0.010310 37 0.457457 0.474462 0.017000 0.806735 0.840000 0.033270 38 0.533847 0.532308 0.001539 0.883930 0.880000 0.003930 39 0.617969 0.619263 0.001290 0.854069 0.840000 0.014069 40 0.675052 0.675232 0.000180 0.854545 0.850000 0.004545 41 0.435924 0.478165 0.042240 0.838353 0.840000 0.001650 42 0.452255 0.484692 0.032440 0.813703 0.850000 0.036300 43 0.739848 0.723855 0.015993 0.938403 0.880000 0.058403 44 0.500706 0.502156 0.001450 0.827064 0.840000 0.012940 45 0.695122 0.494601 0.200521 0.866262 0.860000 0.006262 46 0.738727 0.737289 0.001438 0.848324 0.870000 0.021680 Table V. 47 0.355991 0.355991 1.3 £ 102 8 0.792899 0.800000 0.007100 CCR and BCC efficiency 48 0.662169 0.674932 0.012760 0.800977 0.840000 0.039020 score prediction 49 0.412507 0.435469 0.022960 0.784348 0.840000 0.055650
  • 17. 1.2 Indian business 1 schools 0.8 CCR-EFF Efficiency 0.6 NN-CCR 237 0.4 BCC-EFF 0.2 NN-BCC Figure 4. Comparison of NN 0 predictions with 1 5 9 13 17 21 25 29 33 37 41 45 49 DEA results DMUs In future study, more number of schools over a period of time may be considered for better insight into the problem. References Banker, R.D., Charnes, A. and Cooper, W.W. (1984), “Some models for estimating technical and scale inefficiencies in data envelopment analysis”, Management Science, Vol. 30 No. 9, pp. 1078-92. Charnes, A., Cooper, W.W. and Rhodes, E. (1978), “Measuring the efficiency of decision making units”, European Journal of Operations Research, Vol. 2 No. 6, pp. 429-44. Chiang, W., Urban, T.L. and Baldridge, G.W. (1996), “A neural network approach to mutual fund net asset value forecasting”, Omega, Vol. 24 No. 2, pp. 205-15. Dayal, I. (2002), “Developing management education in India”, Journal of Management Research, Vol. 2 No. 2, pp. 98-113. Dutta, S. and Shekhar, S. (1988), “Bond ratings: a non-conservative application of neural networks”, IEEE International Conference on Neural Networks, San Diego, CA, Vol. 2, pp. 443-50. Hoefer, P. and Gould, J. (2000), “Assessment of admission criteria for predicting students’ academic performance in graduate business programs”, Journal of Education for Business, Vol. 75 No. 4, pp. 225-9. Hu, M.Y., Zhang, G.P. and Haiyang, C. (2004), “Modeling foreign equity control in Sino-foreign joint ventures with neural networks”, European Journal of Operational Research, Vol. 159 No. 3, pp. 729-40. Johnes, G. and Johnes, J. (1993), “Measuring the research performance of UK economics departments: an application of data envelopment analysis”, Oxford Economics Papers, Vol. 4 No. 2, pp. 332-47. Kannan, S.R. (2005), “Extended bidirectional associative memories: a study on poor education”, Mathematical and Computer Modelling, Vol. 42 Nos 3/4, pp. 389-95. Kimoto, T., Asakawa, K., Yoda, M. and Takeoda, M. (1990), “Stock market prediction system with modular neural networks”, Proceedings of the International Joint Conference on Neural Networks, San Diego, CA, Vol. 1, pp. 1-6. Lopes, A.L.M. and Lanzer, E.A. (2002), “Data envelopment analysis – DEA and fuzzy sets to assess the performance of academic departments: a case study at Federal University of Santa Catarina – UFSC”, Pesquisa Operational, Vol. 22 No. 2, pp. 217-30.
  • 18. BIJ Lu, L-C., Chen, W-H., Kim, D. and Hwang, C-P. (1996), “Artificial neural systems improve franchising decision making”, International Journal of Management, Vol. 13 No. 2, pp. 25-32. 18,2 McMillan, L.M. and Datta, D. (1998), “The relative efficiencies of Canadian universities: a DEA perspective”, Canadian Public Policy, Vol. 24 No. 4, pp. 485-511. McMullen, P.R. (1997), “Assessment of MBA programs via data envelopment analysis”, Journal of Business and Management, Vol. 5 No. 1, pp. 77-91. 238 Mintzberg, H. (1973), The Nature of Managerial Work, Harper & Row, New York, NY. Mostafa, M.M. (2009), “Modeling the efficiency of top Arab banks: a DEA-neural network approach”, Expert Systems with Applications, Vol. 36, pp. 309-20. Mozer, M., Wolniewicz, R., Johnson, E. and Kaushansky, H. (1999), “Curn reduction in the wireless industry”, Proceedings of the Neural Information Systems Conference, San Diego, CA. Naik, B. and Ragothaman, S. (2004), “Using neural networks to predict MBA student success”, College Student Journal, Vol. 38, March, pp. 210-8. Nordmann, L.H. and Luxhoj, J.T. (2000), “Neural network forecasting of service problems for aircraft structural component grouping”, Journal of Aircraft, Vol. 37 No. 2, pp. 332-8. Odom, M.D. and Sharda, R. (1990), “A neural network model for bankruptcy prediction”, Proceedings of the International Joint Conference on Neural Networks, San Diego, CA, Vol. 2, pp. 163-8. Ojha, A.K. (2005), “Abhoy management education in India: protecting it from the rankings onslaught”, Decision, Vol. 32 No. 2, pp. 19-33. Palisade Corporation (2008), Neural Tools User Guide Version 5.0, Palisade Corporation, New York, NY. Ramanathan, R. (2001), “A data envelopment analysis of comparative performance of schools in the Netherlands”, Operations Research, Vol. 38 No. 2, pp. 160-81. Ray, C.S. and Jeon, Y. (2003), “Reputation and efficiency: a nonparametric assessment of America’s top-rated MBA programs”, Working Paper 2003-13, March, available at: www. econ.uconn.edu/ (accessed 23 September 2009). Sahay, B.S. and Thakur, R. (2007), “Excellence through accreditation in Indian B-Schools”, Global Journal of Flexible Systems in Management, Vol. 8 No. 4, pp. 9-16. Sahay, B.S. and Thakur, R. (2008), “Making Indian management education globally competitive”, Proceedings of ASBBS, Vol. 15 No. 1, pp. 1332-9. Specht, D.F. (1991), “A general regression neural network”, IEEE Transactions on Neural Networks, Vol. 2 No. 6, pp. 568-76. Sreekumar, S. and Patel, G. (2007), “Comparative analysis of B-school rankings and an alternate ranking method”, International Journal of Operations and Quantitative Management, Vol. 13 No. 1, pp. 33-46. Tam, K.Y. and Kiang, M.Y. (1992), “Managerial applications of neural networks: the case of bank failure predictions”, Management Science, Vol. 38 No. 7, pp. 926-47. Tomkins, C.Y. and Green, R. (1988), “An experiment in the use of data envelopment analysis of evaluating the efficiency of UK university departments of accounting”, Financial Accountability & Management, Vol. 14 No. 2, pp. 147-64. Wadhwa, S., Kumar, A. and Saxena, A. (2005), “Modeling and analysis of technical education system: a KM and DEA based approach”, Studies in Informatics and Control, Vol. 14 No. 4, pp. 235-50. Wang, Z. (1994), “An artificial neural network model for comparative study of education system of China”, Control Engineering Practice, Vol. 2 No. 1, pp. 167-80.
  • 19. Whitley, R., Thomas, A. and Marceau, J. (1981), Masters of Business? Business Schools and Indian business Business Graduates in Britain and France, Tavistock, London. Wu, P., Fang, S-C., King, R.E. and Nuttle, H.L. (1995), “Decision surface modeling of apparel retail schools operations using neural network technology”, International Journal of Operations and Quantitative Management, Vol. 1 No. 1, pp. 33-47. About the authors 239 S. Sreekumar is an Associate Professor in Rourkela Institute of Management Studies, Rourkela 769015, India. His areas of interest include application of DEA for efficiency analysis and multi-criteria decision making. He has 17 years of teaching experience in the areas of quantitative techniques and information science. He has published 30 papers in various international and national journals and conferences. He has also authored two books. S.S. Mahapatra is Professor in the Department of Mechanical Engineering, National Institute of Technology Rourkela, India. He has more than 20 years of experience in teaching and research. His current area of research includes multi-criteria decision making, quality engineering, assembly line balancing, group technology, neural networks, and non-traditional optimization and simulation. He has published more than 40 papers in referred journals. He has written few books related to his research work. He is also currently dealing with few sponsored projects. S.S. Mahapatra is the corresponding author and can be contacted at: mahapatrass2003@yahoo.com To purchase reprints of this article please e-mail: reprints@emeraldinsight.com Or visit our web site for further details: www.emeraldinsight.com/reprints