GitHub - LeoGrin/tabular-benchmark

Accompanying repository for the paper Why do tree-based models still outperform deep learning on tabular data? To download these datasets, simply run python data/download_data.py. We're planning to release a version allowing to use Benchopt instead of WandB to make it easier to run. All the R code used to generate the analyses and figures in available in the analyses folder. The datasets used in the benchmark have been uploaded as OpenML benchmarks, with the same transformations that are used in the paper.