optcat 0.1.0

OptCAT (= Optuna + CatBoost) provides a scikit-learn compatible estimator that tunes hyperparameters in CatBoost with Optuna.

OptCAT

OptCAT (= Optuna + CatBoost) provides a scikit-learn compatible estimator that tunes hyperparameters in CatBoost with Optuna.

This Repository is very influenced by Y-oHr-N/OptGBM.

Examples

from optcat.core import CatBoostClassifier
from sklearn import datasets

params = {
        "bootstrap_type": "Bayesian",
        "loss_function": "Logloss",
        "iterations": 100
    }

model = CatBoostClassifier(params=params, n_trials=5)
data, target = datasets.load_breast_cancer(return_X_y=True)
model.fit(X=data, y=target)

Installation

pip install git+https://github.com/wakamezake/OptCAT.git

Testing

poetry run pytest