I'm needing some advice on whether to use RFE or just simple Elastic Net feature selection. My dataset is a lot wider than it is long, about 7500 variables with ~2500 observations. The 7500 variables are post high correlation removal. I currently am running 3 different models, PLS, SVM, and a KNN model. Currently, I am running RFE Feature selection on each individual model in hopes of optimizing each one. I'm wondering though if I'm wasting my time. The computing time and power to do this is taking a lot and I'm wanting some advice on if it's worth it. One last thing worth mentioning is that my dependent variable can be both positive and negative. Appreciate any help/advice you guys can provide.
Asked
Active
Viewed 47 times
