Metric Random Forest

This repository contains the code that can reproduce results in the paper Medoid splits for efficient random forests in metric spaces (Bulté and Sørensen, 2023).

The implementation of the Metric Random Forest, along with implementation of other regression methods in metric spaces can be found in the dedicated repository: pyfréchet¹. Reproducing the plots can be done by running the notebooks corresponding to each setup. Each notebook contains the necessary code to generate examples shown in the descriptive figures, run the numerical experiments and plot their results.

Footnotes

1. The experiments were conducted based on the pyfrechet version corresponding to the commit ID 78d92cb5f9518b1896df9fb20f966a7d0b4290df ↩