Given we have a binary dependent variable and 100s of features and ~50k observations, is there a generally accepted way to trim the features via some type of machine learning concept? I was trying a Lasso regression to zero out features, but it just showed nothing was significant. I can go through multiple by hand that are definitely significant, though, so I must be doing something wrong. If I had a specific type of selection to look into, I would feel more comfortable learning about that specific concept and knowing it should theoretically work for me.

Sorry I'm a complete noob to this, and am just looking for some general direction.

  • $\begingroup$ Elaborate your question. What is your coding language? Add a snippet of your code so we can take a look at the input and output to give better insights $\endgroup$ – Nain Nov 20 '16 at 10:18

I am not sure if you are using python, but sklearn's documentation has a few good suggestions on commonly used feature selection techniques: http://scikit-learn.org/stable/modules/feature_selection.html

You also may want to look into decomposition techniques such as principal components analysis (PCA) and partial least squares (PLS).


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.