def _objective()

in python/pipelines/components/python/component.py [0:0]

• 16 lines of code
• 4 McCabe index (conditional complexity)

    def _objective(trial):
        # Define the hyperparameter search space.
        # Using Scikit-learn goes beyond using https://cloud.google.com/bigquery/docs/hp-tuning-overview
        params = {
        "n_clusters": trial.suggest_int("n_clusters", min_num_clusters, max_num_clusters),
        "max_iter": trial.suggest_int("max_iter", 10, 1000, step=10),
        "tol": trial.suggest_float("tol", 1e-6, 1e-2, step=1e-6),
        }

        model = _create_model(params)
        model.fit(training_dataset_df)
        labels = model.predict(training_dataset_df)

        return silhouette_score(
            X=model.named_steps['transform'].transform(training_dataset_df),
            labels=labels, 
            metric='euclidean',
            # By default, we're setting sample size to the whole training dataset since we're looking for accurate scores
            # However, if the codes takes a long time to calculate the silhouette score, assuming training size is
            # bigger than 10000, then set a limit on the number of samples to 10000
            sample_size=None if int(len(training_dataset_df)) < 10_000 else 10_000,
            random_state=42
        ), params['n_clusters']