new step on predicative ml

[helenma0223234]

• [x] gen_topk_dummies for selected_cat_large (the threshold should make the final unique values in the col around 100)
• [ ] #25
• [ ] run random forest classifier with outcome variables[1]
• [ ] run naive bayes model
• [ ] get confusion matrix and test accuracy (should be similiar to B_feature_matrix_prep.ipynb)

[rebeccajohnson88]

Things I'd add:

• [x] Make sure to separate into 80% train; 20% test before imputation
• [x] Update script to add in closer to real data: clean/jobs_formod.csv (see slack message)
• [x] I'd avoid onehotencoder and instead use pd.get_dummies --- definitely avoid handcoding and pd.get_dummies should work if you feed it a list of columns
• [ ] i'd separate into two scripts: (1) preprocess that writes four objects: (1) training matrix features; (2) test matrix features; (3) training label; (4) test label and (2) modeling/evaluation script --- the latter should store models in a list and iterate over them