Regression issue
farzane
New Altair Community Member
Hi everybody
I am using linear regression operatore. everything is fine except this issue. I have 15 attributes for linear regression. I use feature selection M5 prime but the final model is weird. I receive "?" instead of numbers in the p-value column. when I remove one attribute form the data, the final result doesn't have "?". Also, when I replace M5 prime with greedy feature selection, the model is different but it doesn't face that problem.
Is there any limitation in linear regression and the number of attributes?
Thank you so much for your favorable responses.
I am using linear regression operatore. everything is fine except this issue. I have 15 attributes for linear regression. I use feature selection M5 prime but the final model is weird. I receive "?" instead of numbers in the p-value column. when I remove one attribute form the data, the final result doesn't have "?". Also, when I replace M5 prime with greedy feature selection, the model is different but it doesn't face that problem.
Is there any limitation in linear regression and the number of attributes?
Thank you so much for your favorable responses.
0
Answers
-
Without seeing your data and your process, it is hard to diagnose. There is no inherent limitation to the number of features usable with M5 prime.
Did you check the parameters regarding removing collinear features? What about adding bias? Both of those can contribute to issues with calculating the p-values.0
