Resampling procedures to assess variable selection stability with finite sample error control, implemented by Benjamin Hofner. The package implements standard stability selection and complementary pairs stability selection, as described in referenced academic papers. It is designed for use with high-dimensional variable selection procedures such as Lasso or boosting.
Use Cases
- Assessing the stability of selected variables based on resampling procedures described.
- Controlling finite sample error for Lasso-based variable selection.
- Applying complementary pairs stability selection with improved error bounds.
- Integrating custom variable selection algorithms with the provided stability framework.
Strengths
- Implements two established methods: standard stability selection and complementary pairs stability selection.
- Designed for integration with arbitrary user-specified variable selection approaches.
- Based on peer-reviewed methodologies from Meinshausen & Buhlmann (2010) and Shah & Samworth (2013).
Limitations
- Description metadata is limited; actual data quality requires manual inspection after download.
- Column-level documentation is absent; field semantics must be inferred after download.
- Row count is unknown, which may limit suitability assessment.
Provenance
- Source
- Benjamin Hofner via paperswithcode
- Collection Method
- Software package implementing statistical resampling methods.