Sklearn interface

[ppdebreuck]

The sklearn API implementation of MODNet can be found under modnet.sklearn. It enables integration with scikit-learn methods such as pipelines, model selection functions (e.g. gridsearch), and integration with other sklearn models.

The main classes are:

• MODNetFeaturizer: A transformer that converts a list of compositions or structures to a dataframe of shape (n_samples, n_features)

• RR: A transformer based on Relevance-Redundancy (RR) feature selection. Given an input array or dataframe it will keep n_feat features with the highest RR-score.

• MODNetRegressor: A regressor based on the MODNetModel for fitting, allowing multiple properties.

• MODNetClassifier: A classifier based on the MODNetModel.

Tests: modnet/tests/test_sklearn.py

Example notebook: here

[davidwaroquiers]

This sounds really nice, thanks a lot! I can review this PR as I initially proposed the feature (without doing the hard work of implementing it ;-) Most likely end of this month.