feature-selector
Development and implementation of the feature selector
Refer to the Feature Selector Usage notebook for how to use
The feature selector is a class for removing features for a dataset intended for machine learning. There are five methods used to identify features to remove:
- Missing Values
- Single Unique Values
- Collinear Features
- Zero Importance Features
- Low Importance Features
The FeatureSelector also includes a number of visualization methods to inspect
characteristics of a dataset.
Requires:
python==3.6+
lightgbm==2.1.1
numpy==1.14.5
pandas==0.23.1
scikit-learn==0.19.1