2. 4 Steps to Input Modeling
1. Collect data from real system
Substantial time and resources
When data is unavailable (due to time
limit or no existing process):
• Use expert opinion
• Make educated guess based from
knowledge of the process
3. 4 Steps to Input Modeling
1. Identify probability distribution to
represent input process
Develop frequency distribution or
histogram
Choose a family of distributions
4. 4 Steps to Input Modeling
1. Choose the parameters of the
distribution family.
These parameters are estimated from the
data.
2. Evaluate the chosen distribution and its
parameters.
Goodness of fit test : chi-square or KS test.
This is an iterative process of selecting and
rejecting the different distributions until the
desired is found.
If none is found, create an empirical
distribution.
5. Data Collection Problems
Inter-arrival times are not
homogenous
Service times which are dependent
on other factors
Service time termination
Machine breakdowns
6. No DataNo Data
Old DataOld Data
Missing DataMissing Data
GuesstimatesGuesstimates
Erroneous DataErroneous Data
No resourceNo resource
Data Problems with
Simulation
SimplifyingSimplifying
assumptionsassumptions
Using AveragesUsing Averages
OutliersOutliers
Optimistic DataOptimistic Data
PoliticsPolitics
7. Bad data equals bad modelsBad data equals bad models
The Best models fail under badThe Best models fail under bad
datadata
Successful simulation is unlikelySuccessful simulation is unlikely
with bad datawith bad data
Consequence of Data Problems
8. Always question dataAlways question data
Electronic data does mean goodElectronic data does mean good
data.data.
Know the sourceKnow the source
Allocate sufficient time to collectAllocate sufficient time to collect
and analyze dataand analyze data
Guidelines in Data
Collection
9. Suggestions to facilitate
data collection:
1. Plan
Collect data while pre-observing
Create forms and be prepared to
modify them when needed
Video tape is possible and extract date
later
10. Suggestions to facilitate
data collection:
1. Analyze.
Determine if data is adequate.
Do not collect superfluous data.
2. Try to combine homogenous data.
Use two sample t-test.
3. Be wary of data censoring.
4. Look for relationships between variables
using a scatter plot.
5. Be aware of autocorrelations within a
sequence of observations.