Gradient Boosting Mapping is and nonlinear dimensionality reduction and feature creation method. This repository contains the Python code for GBMAP and the experiments in the paper:
Accepted to Discovery Science 2024, meanwhile check out a previous preprint http://arxiv.org/abs/2405.08486.
Most of the datasets are from OpenML and will be downloaded as needed, but GeckoQ has to be downloaded separately.
The GeckoQ data can be downloaded from Fairdata repository, you'll only have to download the Dataframe.csv and place it to experiments/data.
Script run.sh contains explicit instructions how to install and run the experiments. The results for the experiments are placed to experiments/results and figures to experiments/figures.