Handling of Discrete and Continuous variables - Two kernel types

transferwise / hisel

Feature selection tool based on Hilbert-Schmidt Independence Criterion

Apache License 2.0

2 stars 0 forks source link

Handling of Discrete and Continuous variables - Two kernel types #40

Closed claudio-tw closed 1 year ago

claudio-tw commented 1 year ago

Context

This PR implements an additional feature of hisel. More precisely, we add the capability of using two different kernels when computing the Gram matrix. This allows us to handle with HSIC Lasso the case where the features to select from are both categorical and continuous.

Checklist

[x] Change meets or does not compromise the Baseline Security Requirements