Example lgbm terminator

lightgbm/lightgbm_terminator.py

@@ -0,0 +1,55 @@
+"""
+Optuna example showcasing the new Optuna Terminator feature.
+
+In this example, we utilize the Optuna Terminator for hyperparameter
+optimization on a lightgbm using the wine dataset.
+The Terminator automatically stops the optimization process based
+on the potential for further improvement.
+
+To run this example:
+
+    $ python lightgbm_terminator.py
+
+"""
+
+import optuna
+from optuna.terminator.callback import TerminatorCallback
+from optuna.terminator.erroreval import report_cross_validation_scores
+
+from sklearn.datasets import load_wine
+import lightgbm as lgb
+from sklearn.model_selection import cross_val_score
+from sklearn.model_selection import KFold
+
+
+def objective(trial):
+    X, y = load_wine(return_X_y=True)
+
+    params = {
+        "objective": "multiclass",
+        "num_class": 3,
+        "verbosity": -1,
+        "boosting_type": "gbdt",
+        "lambda_l1": trial.suggest_float("lambda_l1", 1e-8, 10.0, log=True),
+        "lambda_l2": trial.suggest_float("lambda_l2", 1e-8, 10.0, log=True),
+        "num_leaves": trial.suggest_int("num_leaves", 2, 256),
+        "feature_fraction": trial.suggest_float("feature_fraction", 0.4, 1.0),
+        "bagging_fraction": trial.suggest_float("bagging_fraction", 0.4, 1.0),
+        "bagging_freq": trial.suggest_int("bagging_freq", 1, 7),
+        "min_child_samples": trial.suggest_int("min_child_samples", 5, 100),
+    }
+
+    clf = lgb.LGBMClassifier(**params)
+
+    scores = cross_val_score(clf, X, y, cv=KFold(n_splits=5, shuffle=True))
+    report_cross_validation_scores(trial, scores)
+
+    return scores.mean()
+
+
+if __name__ == "__main__":
+    study = optuna.create_study(direction="maximize")
+    study.optimize(objective, n_trials=50, callbacks=[TerminatorCallback()])
+
+    print(f"The number of trials: {len(study.trials)}")
+    print(f"Best value: {study.best_value} (params: {study.best_params})")

lightgbm/requirements.txt

@@ -2,3 +2,4 @@ lightgbm>=3.3.0
 numpy
 optuna
 scikit-learn
+botorch

terminator/requirements.txt

@@ -1,3 +1,5 @@
 botorch
 scikit-learn
 plotly
+lightgbm
+optuna

terminator/terminator_lgbm.py

@@ -0,0 +1,55 @@
+"""
+Optuna example showcasing the new Optuna Terminator feature.
+
+In this example, we utilize the Optuna Terminator for hyperparameter
+optimization on a lighgbm using the wine dataset.
+The Terminator automatically stops the optimization process based
+on the potential for further improvement.
+
+To run this example:
+
+    $ python terminator_lgbm.py
+
+"""
+
+import optuna
+from optuna.terminator.callback import TerminatorCallback
+from optuna.terminator.erroreval import report_cross_validation_scores
+
+from sklearn.datasets import load_wine
+import lightgbm as lgb
+from sklearn.model_selection import cross_val_score
+from sklearn.model_selection import KFold
+
+
+def objective(trial):
+    X, y = load_wine(return_X_y=True)
+
+    params = {
+        'objective': 'multiclass',   
+        'num_class': 3,  
+        "verbosity": -1,
+        "boosting_type": "gbdt",
+        "lambda_l1": trial.suggest_float("lambda_l1", 1e-8, 10.0, log=True),
+        "lambda_l2": trial.suggest_float("lambda_l2", 1e-8, 10.0, log=True),
+        "num_leaves": trial.suggest_int("num_leaves", 2, 256),
+        "feature_fraction": trial.suggest_float("feature_fraction", 0.4, 1.0),
+        "bagging_fraction": trial.suggest_float("bagging_fraction", 0.4, 1.0),
+        "bagging_freq": trial.suggest_int("bagging_freq", 1, 7),
+        "min_child_samples": trial.suggest_int("min_child_samples", 5, 100),
+    }
+
+    clf = lgb.LGBMClassifier(**params)
+
+    scores = cross_val_score(clf, X, y, cv=KFold(n_splits=5, shuffle=True))
+    report_cross_validation_scores(trial, scores)
+
+    return scores.mean()
+
+
+if __name__ == "__main__":
+    study = optuna.create_study(direction="maximize")
+    study.optimize(objective, n_trials=50, callbacks=[TerminatorCallback()])
+
+    print(f"The number of trials: {len(study.trials)}")
+    print(f"Best value: {study.best_value} (params: {study.best_params})")