rsteca / sklearn-deap

Use evolutionary algorithms instead of gridsearch in scikit-learn
MIT License
771 stars 131 forks source link

Added hall_of_fame attribute such that users can retrieve the k best … #41

Open cfusting opened 6 years ago

cfusting commented 6 years ago

…scores and parameters from a run.

ryanpeach commented 6 years ago

Huh, ok. I thought for sure we had one of those. It's a little redundant since we have the entire history, logbook, best_scores and best_parameters returned, but if you want the deap native implementation I don't have a problem with it. Please duplicate all your work with the maximize function on the optimize.py file, and make sure it passes (or else change) the test.py script, and your good.

ryanpeach commented 6 years ago

if you have to add another output to maximize, please change it so that it returns an extra dict instead so we can start simply naming additional new outputs.

cfusting commented 6 years ago

I felt like using the other objects to retrieve the hall of fame was a bit circuitous. I'll add the item to maximize.

cfusting commented 6 years ago

This is all ready to go.

rsteca commented 6 years ago

Can you fix the conflicts so I can merge this? Thanks for the pull request!

cfusting commented 6 years ago

I resolved these as best I could but I think another commit occurred around the same time as mine. Perhaps we should have the other merger have a look over my resolutions.