I decided to use RFE using the caret package for feature selection for a logistic regression model.

The documentation says the Varimp for linear model uses

the absolute value of the t-statistic for each model parameter is used.

Logistic regression is not a linear model or has any of the linear model assumptions. Does it make sense to use linear model assumptions to reduce variables for a logistic regression problem using the caret package?


We wouldn't be making those assumptions. The logistic regression model falls into a wider class called generalized linear models (as does linear regression).

The t-test discussed here is the generalized linear model t-statistic to test that the parameter is equal to zero.

  • $\begingroup$ Max can i use this for a neural network model as well ? $\endgroup$ – Stat question Nov 6 '13 at 17:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.