chimeraboost
Lightning-fast Gradient Boosting, near-CatBoost quality, all in Python
📖 Documentation: bbstats.github.io/chimeraboost
Install
pip install chimeraboost
[OPTIONAL]:
chimeraboost-warmup
chimeraboost-warmup compiles the numba kernels once and caches them, so the first
fit is not several seconds slower than the rest. Re-run it after every upgrade, which
resets the cache. See
Deployment.
Quickstart
from chimeraboost import ChimeraBoostClassifier, ChimeraBoostRegressor
# classification. quality picks the speed/accuracy trade-off: 1 fastest .. 5 strongest,
# defaulting to 3.
clf = ChimeraBoostClassifier(quality=5)
clf.fit(X, y, cat_features=[0, 1], sample_weight=w)
proba = clf.predict_proba(X_test)
# regression (RMSE, MAE, Quantile, Huber, Poisson, Gamma, Tweedie, or your own)
reg = ChimeraBoostRegressor(loss="Quantile", alpha=0.9)
reg.fit(X, y)
What it is
An opinionated GBDT library that only depends on common Python libraries (NumPy, numba, scikit-learn, SciPy):
- Regression, quantile regression, binary and multiclass classification
What makes it different?
- Bagging as a first-class feature (
n_ensembles) - Automatic early stopping
- Fast multi-quantile fitting
(
ChimeraBoostQuantileRegressor) - Automatic linear-leaf auditioning
- Exact SHAP explanations (
model.shap_values(X)).
Scored against CatBoost and LightGBM only; how the suite is built and weighted is in docs/benchmarks.md. On the sealed TabArena leaderboard the default scores above XGBoost and LightGBM while training faster than either; CatBoost scores higher and takes considerably longer (chart).
Documentation
- Getting started: install and first models
- Recipes: categoricals, quantiles, bagging, custom losses, and more
- Parameters: every option, with defaults and guidance
- FAQ: common questions
Why?
- I want to be able to modify my GBDT library at will
- I know Python and I don't know C
Inspirations / Citations
- CatBoost, Prokhorenkova et al., NeurIPS 2018. Ordered boosting, ordered target statistics, oblivious trees.
- XGBoost, Chen & Guestrin, KDD 2016. Regularized objective, Newton leaf estimation, column subsampling.
- LightGBM, Ke et al., NeurIPS 2017. Histogram-based split finding.
- Linear-leaf trees, Shi et al., IJCAI 2019 (arXiv:1802.05640). Piece-wise-linear regression trees (
linear_leaves). - TreeSHAP, Lundberg et al., Nature Machine Intelligence 2020 (orig. SHAP, NeurIPS 2017). Exact additive feature attributions (
shap_values). - OpenFE, Zhang et al., ICML 2023 (arXiv:2211.12507). Automated pairwise feature generation (
cross_features). - Conformalized quantile regression, Romano, Patterson & Candès, NeurIPS 2019. Distribution-free interval calibration (
conformalize). - TabArena, Erickson et al., NeurIPS 2025 (arXiv:2506.16791). The tabular benchmark used for evaluation.
1 maintainer
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file chimeraboost-0.28.0.tar.gz.
File metadata
- Download URL: chimeraboost-0.28.0.tar.gz
- Upload date:
- Size: 229.9 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via:
twine/6.2.0 CPython/3.13.13
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
f1707618582cfecbe9913e49a53effed7c63541a06d3c01a4ab6ab71067fafc2
|
|
| MD5 |
c9bde3fad4543c896bc2398b74c5358c
|
|
| BLAKE2b-256 |
77852149efe417bd3025348763e48de3e3eefc2b9eca4d1a547fe527b47c90d9
|
File details
Details for the file chimeraboost-0.28.0-py3-none-any.whl.
File metadata
- Download URL: chimeraboost-0.28.0-py3-none-any.whl
- Upload date:
- Size: 135.9 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via:
twine/6.2.0 CPython/3.13.13
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
00eb95cc9604d9fa2f08a5b35989b9d3634262dba4cbdfcadaf8948bda752f5f
|
|
| MD5 |
535435ad95b47a7276a4b9895febd5f8
|
|
| BLAKE2b-256 |
e76b8c9b3d96f778316ef31acb6948ea28b57a6ae95abf2d0cec65b2d158a0de
|
