fastfrechet: An R package for fast implementation of Fréchet regression with distributional responses.

Area of Science:

• Statistical modeling
• Computational statistics
• Biomedical data analysis

Background:

• Distribution-as-response regression is increasingly important in biomedical fields with complex patient data.
• Fréchet regression handles non-Euclidean and distributional responses, extending traditional regression.
• Existing methods for variable selection in this domain are being advanced.

Purpose of the Study:

• To introduce fastfrechet, an R package for efficient Fréchet regression and variable selection.
• To provide a scalable implementation for analyzing large, distribution-valued biomedical datasets.
• To make advanced statistical methods readily accessible for researchers.

Main Methods:

• Implementation of Fréchet regression and variable selection algorithms in R.
• Utilizes the 2-Wasserstein metric for distributional responses.
• Incorporates resampling tools for automated variable selection.

Main Results:

• The fastfrechet package offers a fast and scalable solution for distribution-based regression.
• Enables efficient variable selection using resampling techniques.
• Facilitates the analysis of large-scale biomedical datasets like the UK Biobank.

Conclusions:

• fastfrechet democratizes the use of advanced Fréchet regression and variable selection methods.
• The package is highly scalable and suitable for large biomedical data.
• It supports robust analysis of complex, distribution-valued data.