Feature selection in online learning fashion

[Haoyu-R]

Hi there,

I just came up with an idea: we can implement some online feature selection methods to eliminate/exclude some irrelevant features before calling the real online-ML algorithm. Also, feature selection can be used as an element in the pipeline of the river. This will improve the overall efficiency of online-ML.

Some references:
https://arxiv.org/abs/1511.09263
https://www.sciencedirect.com/science/article/abs/pii/S0950705117304124

Cheers.

[MaxHalford]

Hey! We've already started a feature_selection module, see here. But feel free to improve it and/or contribute more methods!