Help with features reduction and combining features

[Isabelle-Simard]

@illdopejake I might need your help next week to learn how to do some features reduction when I am going to combine my features and also how exactly to combine the predictive features in the same model.

[atrophiedbrain]

I would be interested in sitting in on this meeting, if possible. I may have something useful to add, but more than likely it will be a learning opportunity for me as well.

[illdopejake]

Perfect! Let's organize it on slack. As of now, I will be around on Wednesday and Thursday next week, but you can always look at the calendar for up-to-date schedule of instructors!

Until then, there are some resources you can look into to get you started on concepts and trying out strategies. If your aim is dimension reduction (unsupervised feature reduction), you can get started with some of these functions, and here is an example of a slightly more in-depth article on the concepts (though you can find others online).

Regarding feature selection (supervised feature reduction), sklearn has a nice short sweet description of approaches in their documentation. Or, here is a really nice article on the topic.

With a smaller sample size, I think dimension reduction is the way to go -- feature selection can be tricky to use effectively. But anyway, you can try out a few things and maybe we can go over it next week! Good luck, and don't be afraid to reach out with other questions :-)

[Isabelle-Simard]

Thanks so much Jake! This is all very helpful!