|
|
 |
| Ratio of correct prediction |
 
|
| Message/Author |
|
|
| TC Chiang posted on Friday, November 11, 2016 - 10:50 am
|
|
|
Hi,
I used Mplus to solve simultaneous logistic regression models as the follows,
I want to know if and how I can get the ratio of correct prediction of the models, in addition to the goodness of fit of the model. (I have read through the Mplus manuals but failed to work it out).
i.e., I want to know how well the models predict, in terms of percentage of correct prediction.
Please assist it and many many thanks in advance.
-----------------------
Title:
Simutanous probit regression
Data:
File is "D:/xxx.csv";
Variable:
Names are ONUM SIC BRAND TBRS NATION SERV SIZE CON ADVI ANUM
TADV AADV RADV STC RNDI TEFCY IRATIO EDM IDC VALUE LABOR
ENGY GDP CDIS QUNC SPRO AINT SIND LITY EITY STBB RADV100th;
CATEGORICAL ARE NATION BRAND;
USEVARIABLES = NATION LITY EITY AINT TEFCY TBRS BRAND
RADV100th SPRO QUNC SIND STBB;
Analysis:
estimator = WLSMV; !ML;
parameterization = theta;
Model:
NATION on LITY EITY AINT TEFCY TBRS BRAND;
BRAND on RADV100th SPRO QUNC SIND STBB NATION; |
|
|
|
|
Percentage correct prediction can be obtained by comparing the estimated probability for a subject with the observed 0/1 outcome. How to get the estimated probability is shown on our book page
http://www.statmodel.com/mplusbook/chapter5.shtml
See the input/output for 5.11 (Part 2) and 5.11 (Part 3).
There is no generally accepted overall goodness of fit measure for logistic regression. |
|
|
| TC Chiang posted on Friday, November 11, 2016 - 1:16 pm
|
|
|
Hi Bengt,
Thanks for reply. I tested and found the accuracy of predicted values is 1. That means the "propensity" output/save is not the estimated values by the model, but the original data.
Did I miss something? Look forward to hearing from you soon. Thanks again. |
|
|
|
|
| Propensity is the estimated probability. If this doesn't help, send relevant files to Support along with your license number. |
|
|
| Back to top |
|
|
