PyTabKit: Tabular ML models and benchmarking

Paper	Documentation	RealMLP-TD-S standalone implementation	Grinsztajn et al. benchmark code	Data archive

PyTabKit provides scikit-learn interfaces for modern tabular classification and regression methods benchmarked in our paper, see below. It also contains the code we used for benchmarking these methods on our meta-train and meta-test benchmarks.

Installation

pip install pytabkit

If you want to use TabR, you have to manually install faiss, which is only available on conda.
You might need to install swig (e.g. via pip) if the build of pyrfr fails.
Please install torch separately if you want to control the version (CPU/GPU etc.)
Use pytabkit[full] to also install the benchmarking library part. See also the documentation. To run the data download, you need one of rar, unrar, or 7-zip to be installed on the system.

Using the ML models

Most of our machine learning models are directly available via scikit-learn interfaces. For example, you can use RealMLP-TD for classification as follows:

from pytabkit.models.sklearn.sklearn_interfaces import RealMLP_TD_Classifier

model = RealMLP_TD_Classifier()  # or TabR_S_D_Classifier, CatBoost_TD_Classifier, etc.
model.fit(X_train, y_train)
model.predict(X_test)

The code above will automatically select a GPU if available, try to detect categorical columns in dataframes, preprocess numerical variables and regression targets (no standardization required), and use a training-validation split for early stopping. All of this (and much more) can be configured through the constructor and the parameters of the fit() method. For example, it is possible to do bagging (ensembling of models on 5-fold cross-validation) simply by passing n_cv=5 to the constructor. Here is an example for some of the parameters that can be set explicitly:

from pytabkit.models.sklearn.sklearn_interfaces import RealMLP_TD_Classifier

model = RealMLP_TD_Classifier(device='cpu', random_state=0, n_cv=1, n_refit=0,
                              verbosity=2, val_metric_name='cross_entropy',
                              n_epochs=256, batch_size=256, hidden_sizes=[256] * 3,
                              lr=0.04, use_ls=False)
model.fit(X_train, y_train, val_idxs=val_idxs, cat_features=cat_features)
model.predict_proba(X_test)

See this notebook for more examples. Missing numerical values are currently not allowed and need to be imputed beforehand.

Available ML models

Our ML models are available in up to three variants, all with best-epoch selection:

library defaults (D)
our tuned defaults (TD)
random search hyperparameter optimization (HPO), sometimes also tree parzen estimator (HPO-TPE)

We provide the following ML models:

RealMLP (TD, HPO): Our new neural net models with tuned defaults (TD) or random search hyperparameter optimization (HPO)
XGB, LGBM, CatBoost (D, TD, HPO, HPO-TPE): Interfaces for gradient-boosted tree libraries XGBoost, LightGBM, CatBoost
MLP, ResNet (D, HPO): Models from Revisiting Deep Learning Models for Tabular Data
TabR-S (D): TabR model from TabR: Tabular Deep Learning Meets Nearest Neighbors
Ensemble-TD: Weighted ensemble of all TD models (RealMLP, XGB, LGBM, CatBoost)

Benchmarking code

Our benchmarking code has functionality for

dataset download
running methods highly parallel on single-node/multi-node/multi-GPU hardware, with automatic scheduling and trying to respect RAM constraints
analyzing/plotting results

For more details, we refer to the documentation.

Citation

If you use this repository for research purposes, please cite our paper:

@article{holzmuller2024better,
  title={Better by default: Strong pre-tuned MLPs and boosted trees on tabular data},
  author={Holzm{\"u}ller, David and Grinsztajn, Leo and Steinwart, Ingo},
  journal={arXiv preprint arXiv:2407.04491},
  year={2024}
}

Contributors

David Holzmüller (main developer)
Léo Grinsztajn (deep learning baselines, plotting)
Ingo Steinwart (UCI dataset download)
Katharina Strecker (PyTorch-Lightning interface)
Jérôme Dockès (deployment, continuous integration)

Acknowledgements

Code from other repositories is acknowledged as well as possible in code comments. Especially, we used code from https://github.com/yandex-research/rtdl and sub-packages (Apache 2.0 license), code from https://github.com/catboost/benchmarks/ (Apache 2.0 license), and https://docs.ray.io/en/latest/cluster/vms/user-guides/community/slurm.html (Apache 2.0 license).

dholzmueller / pytabkit