mboukabous's picture
Add application file
7c045bd
raw
history blame contribute delete
717 Bytes
"""
CatBoost Classifier setup.
Features:
- Uses `CatBoostClassifier`.
- Handles categorical features natively but we still rely on pipeline encoding.
- Good for both binary and multi-class.
- Default scoring: 'accuracy'.
Requires `catboost` installed.
"""
from catboost import CatBoostClassifier
estimator = CatBoostClassifier(verbose=0, random_state=42)
param_grid = {
'model__iterations': [100],
'model__depth': [3, 5],
'model__learning_rate': [0.01, 0.1],
# Preprocessing params
#'preprocessor__num__imputer__strategy': ['mean','median'],
#'preprocessor__num__scaler__with_mean': [True,False],
#'preprocessor__num__scaler__with_std': [True,False],
}
default_scoring = 'accuracy'