4 ms·
This is great and has an awesome level of detail on information gain. I have found decision trees really useful for getting a feel for what features in a datase
by pcprincipal 8y ago
This is great and has an awesome level of detail on information gain. I have found decision trees really useful for getting a feel for what features in a dataset matter and trying out different max depths on trees to get more insight into the data.
I wrote an article earlier this year on how I use decision trees to classify players for daily fantasy sports into different groups that people may find useful https://medium.com/@bmb21/why-is-caris-levert-projected-for-53-points-a-decisiontreeregressorstory-deugging-story-c6071ee44efb https://medium.com/@bmb21/why-is-caris-levert-projected-for-...
- ishansharma 8y agoThank you for the kind words. My goal was to make the explanations as simple as possible. Using them to get a feel about dataset is an interesting use case and I haven't tried that. It does sound interesting though. Something to do with my next ML problem I suppose. :) Just skimmed through your article(class final exams tomorrow), it is informative. Thanks for sharing.
- claytonjy 8y agoI use RF's commonly for getting that same feel, but I recently learned that I've been making some big mistakes when interpreting default feature-importance outputs; this recent article really opened my eyes: http://parrt.cs.usfca.edu/doc/rf-importance/index.html http://parrt.cs.usfca.edu/doc/rf-importance/index.html Short version: always use permutation importance, but use leave-it-out importance when it really matters