LassoBacktracking: Implementation for High-Dimensional Data with Feature Interactions

An implementation of the algorithm from Shah (2016) for modeling interactions in high-dimensional data. The algorithm performs sequential Lasso fits on design matrices containing increasing sets of candidate interactions. It was authored by Rajen D. Shah of the University of Cambridge.

Use Cases

• Modeling feature interactions in high-dimensional data based on the described algorithm.
• Performing efficient sequential Lasso regression on large sets of candidate interactions.
• Analyzing datasets with thousands of predictors where interaction effects are important.

Strengths

• The algorithm is described as very efficient, using previous fits to speed up subsequent ones.
• The implementation can handle data with thousands of predictors.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Last update date is unknown; freshness unverified.

Provenance

Source

University of Cambridge

Collection Method

Algorithm implementation based on a 2016 JMLR paper.