I have a logistic regression model that I am working on for a school project. I have about 55 predictors, all of which are continuous. I am relatively new to the idea of "interactions" between variables. What is the process for choosing interactions between variables and then adding them into a logistic regression model?

  • 1
    $\begingroup$ Could you be more specific concerning what you would view as a "simple explanation"? Have you searched our site for other threads about interpreting interactions? I can find several hundred that appear to answer different versions of your question, depending on the type of regression and the nature of the independent variables. $\endgroup$ – whuber Jan 11 at 21:54
  • $\begingroup$ Okay I updated the post, Hopefully this is more clear, but I was curious about interactions in general. Thank you. $\endgroup$ – PastaHo69 Jan 11 at 22:08
  • 2
    $\begingroup$ This really depends on the purpose of your regression. If you're only interested in purely predicting and do not care about interpreting the coefficients or your model more generally, one could select interactions that provided the best predictions as judged by something like out-of-sample error. Otherwise, you should choose your interactions more carefully. Several procedures exist for doing so, including my person favorite called purposeful selection of variables. Search CV for "Purposeful selection" or my answer here: bit.ly/2AIKd1G for one method. $\endgroup$ – StatsStudent Jan 12 at 0:41

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Browse other questions tagged or ask your own question.