Update the cross-validation to be able to use user-inputted groups

civisanalytics / python-glmnet

A python port of the glmnet package for fitting generalized linear models via penalized maximum likelihood.

Other

262 stars 59 forks source link

Closed mwjin closed 4 years ago

mwjin commented 4 years ago

Linear and logistic regressions were updated to be able to use user-inputted labels when splitting training sets for cross validation.
The groups argument is equivalent to the foldid argument of cvglmnet in R.
Use joblib and six modules instead of sklearn.external.
Check the validity of the shape of weights in logitnet.py.

mheilman commented 4 years ago

thanks for the PR!

mheilman commented 4 years ago

Just a note: this is related to #61 since joblib was updated here. A new release is still needed to get these changes out.