The learning outcomes of this topic are:
- Evaluate results from regression analysis
- Interpret results from regression analysis
- Recognise the possibility to extend regression analysis (dummy variables)
2. This topic will cover:
◦ Regression using software
◦ Multiple linear regression
Extending regression models with dummy
variables
Interpreting models
3. By the end of this topic students will be able
to:
◦ Evaluate results from regression analysis
◦ Interpret results from regression analysis
◦ Recognise the possibility to extend regression
analysis
Dummy variables
4. • For the least SSE straight line,
y = mx + c
m =
n xy − x y
n x2 − x 2
c = y − mx
R =
n xy− x y
n x − x 2 n y2 − y 2
5. x y
25 1.44
50 5.58
75
14.6
4
100 6.94
x y
25 3.07
50 5.64
75 9.63
100
10.2
6
6. ◦ Many software packages
OpenOffice, Gretl
MS Excel
SPSS, Minitab, SAS
R, S
◦ MS Excel
2007/2010 Data/Data Analysis/Regression
Older Tools/Data Analysis/ Regression
Then all versions are similar
7.
8.
9. ◦ Reasons Not to Set Constant Term to Zero
It prevents model being biased
Usually interested in the predictor variables anyway
You don’t need to collect data for it
Can help if data is only locally linear
◦ Reason to Set constant Term to Zero
If it is supposed to be zero
Strong theoretical grounds
But care needed with calculation and
interpretation of R2
10.
11. ◦ Models such as
y = c + b1x1+ b2x2+ ...
◦ How are they developed?
an expert task
◦ Managers
understand
question
use results
12. ◦ Estate agent (realtor) is establishing an office in a
new location
◦ Wishes to build a model of advertised prices
◦ Collects competitor data on;
Internal area
Land
Distance from nearest school
City region
13. Property Price School Land Area District
1 457 3 1791 165 FD
2 487 1 800 177 FD
3 218 3 759 94 FD
4 300 4 829 137 FD
5 358 2 630 110 AC
6 658 1 655 201 AC
7 402 2 999 85 AC
8 541 2 920 146 AC
9 358 3 1185 112 Other
10 444 1 787 155 Other
11 298 3 597 180 Other
12 462 1 1447 200 Other
14. Property Price School Land Area AC FD District
1 457 3 1791 165 0 1 FD
2 487 1 800 177 0 1 FD
3 218 3 759 94 0 1 FD
4 300 4 829 137 0 1 FD
5 358 2 630 110 1 0 AC
6 658 1 655 201 1 0 AC
7 402 2 999 85 1 0 AC
8 541 2 920 146 1 0 AC
9 358 3 1185 112 0 0 Other
10 444 1 787 155 0 0 Other
11 298 3 597 180 0 0 Other
12 462 1 1447 200 0 0 Other
15. price =
constant +
a x (km from a school) +
b x (land in m2) +
g x (floor area in m2) +
d (if in AC) +
z (if in FD)
19. R2 =
y − y 2
y − y 2
R
2
= 1- (1 - R2)
n −1
n − k − 1
20.
21.
22.
23.
24. ◦ Equation
expected price = 45.72 + (2.179 x area) +
(148.8 x AC)
◦ Suppose property is in AC and is of 100m2 what is
expected advertised price?
expected price = 45.72 + (2.179 x 100) +
(148.8 x 1)
expected price = 412.42
25. By the end of this topic students will be able
to:
◦ Evaluate results from regression analysis
◦ Interpret results from regression analysis
◦ Recognize possibility to extend regression analysis
Dummy variables
26. ◦ Hinton, PR. Statistics Explained. Routledge
◦ Keast, S. and Towler M. Rational Decision Making
for Managers. Wiley
◦ Wisniewski, M. Quantitative Methods for Decision
Makers. FT Prentice Hall