topic Re: Regression Analysis in Statistical Procedures

Regression Analysis

KafeelBasha — Fri, 03 Jun 2016 06:28:27 GMT

Hello

In logistic regression as we split data into Training and Validation Samples.

Can we do the same in Linear regression?. If not please let me know why?.

Kafeel

Reeza — Fri, 03 Jun 2016 08:04:18 GMT

Yes, you can and should split your data for linear regression.

This methodology is consistent across various statistical procedures not just

logistic regression.

I'm not aware of an automatic method to factor in the diff datasets but I'd review the documentation.

Ksharp — Fri, 03 Jun 2016 09:00:39 GMT

Yes. You could. But I would not recommend .

PROC REG is for predicted model ( interpolated data) , not for forecasted model (Extrapolated data) -- which is well-knowed as Time Series Analysis .

If you use PROC REG for Extrapolated data, the model error would be very large, and model would not be trusted .