Hello Richard,
I am trying to generate a prediction-classification table. It shows how the
model predicts the binary dependent correctly. The cut-off value is usually
50%. For example, EViews produces the following table :
Dependent Variable: PW
Method: ML - Binary Probit (Quadratic hill climbing)
Date: 11/22/06 Time: 14:32
Sample (adjusted): 1 266
Included observations: 266 after adjustments
Prediction Evaluation (success cutoff C = 0.5)
Estimated Equation
Constant Probability
Dep=0
Dep=1
Total
Dep=0
Dep=1
Total
P(Dep=1)<=C
230
27
257
235
31
266
P(Dep=1)>C
5
4
9
0
0
0
Total
235
31
266
235
31
266
Correct
230
4
234
235
0
235
% Correct
97.87
12.90
87.97
100.00
0.00
88.35
% Incorrect
2.13
87.10
12.03
0.00
100.00
11.65
Total Gain*
-2.13
12.90
-0.38
Percent Gain**
NA
12.90
-3.23
Estimated Equation
Constant Probability
Dep=0
Dep=1
Total
Dep=0
Dep=1
Total
E(# of Dep=0)
212.59
22.71
235.31
207.61
27.39
235.00
E(# of Dep=1)
22.41
8.29
30.69
27.39
3.61
31.00
Total
235.00
31.00
266.00
235.00
31.00
266.00
Correct
212.59
8.29
220.88
207.61
3.61
211.23
% Correct
90.46
26.73
83.04
88.35
11.65
79.41
% Incorrect
9.54
73.27
16.96
11.65
88.35
20.59
Total Gain*
2.12
15.07
3.63
Percent Gain**
18.18
17.06
17.62
*Change in "% Correct" from default (constant probability)
specification
**Percent of incorrect (default) prediction corrected by equation
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
