Remove unscreened GLM
jrgant opened this issue · comments
The pattern matching process for grabbing the learners in fhp_spec_learners() grabs the unscreened GLM learner. Only intended to use the screened version because we expect the former to be overfit.
Preliminary are currently running which include this learner. I don't believe it should affect the ensemble learner, but I should check when those jobs get run again for the final numbers.
Edited this function so that the initial glm and screener learners are contained only inside the function environment. lrnr_screen_glm, therefore is the only one export to .GlobalEnv.