lgbm_custom_metric

import numpy as np
import pandas as pd
from lightgbm import LGBMRegressor, log_evaluation
from sklearn.datasets import load_breast_cancer
from sklearn.metrics import mean_absolute_error
from sklearn.model_selection import KFold

data = load_breast_cancer()
X = pd.DataFrame(data.data, columns=data.feature_names)
y = pd.Series(data.target)

folds = KFold(5)


def neg_correlation(preds, labels):
    is_higher_better = False
    return "neg_correlation", -np.corrcoef(preds, labels)[1, 0], is_higher_better


eval_results = []
for train_idx, val_idx in folds.split(X):
    X_train, X_valid = X.iloc[train_idx], X.iloc[val_idx]
    y_train, y_valid = y.iloc[train_idx], y.iloc[val_idx]
    model = LGBMRegressor(n_estimators=100)
    model.fit(
        X.iloc[train_idx],
        y.iloc[train_idx],
        eval_set=[(X.iloc[val_idx], y.iloc[val_idx])],
        eval_metric=neg_correlation,
        callbacks=[log_evaluation(0)],
    )
    eval_results.append(
        np.asarray(model.evals_result_["valid_0"]["neg_correlation"])[:, np.newaxis]
    )

cv_results = np.hstack(eval_results)
best_n_estimators = np.argmin(cv_results.mean(axis=1)) + 1

model = LGBMRegressor(n_estimators=best_n_estimators)
model.fit(X, y, callbacks=[log_evaluation(0)])