si4pipeline

This package provides the statistical test for any feature selection pipeline by selective inference. The tequnical details are described in the paper "Statistical Test for Feature Selection Pipelines by Selective Inference".

Installation & Requirements

This package has the following dependencies:

• Python (version 3.10 or higher, we use 3.12.5)
  • numpy (version 1.26.4 or higher but lower than 2.0.0, we use 1.26.4)
  • scikit-learn (version 1.5.1 or higher, we use 1.5.1)
  • sicore (version 2.3.0 or higher, we use 2.3.0)
  • tqdm (version 4.66.5 or higher, we use 4.66.5)

To install this package, please run the following commands (dependencies will be installed automatically):

$ pip install si4pipeline

Usage

The implementation we developed can be interactively executed using the provided demonstration.ipynb file in our repository. This file contains a step-by-step guide on how to use the package, how to construct a feature selection pipeline, and how to apply the proposed method to a given feature selection pipeline.