List of Built-in Metrics

This package provides some commonly used metrics. They are implemented in a pre-defined protocols in order to be called by experiments classes. Here is an example.

def accuracy_score(y_true, y_pred, param_dict=None):
    """Accuracy classification score.

    Parameters
    ----------
    y_true : 1d array-like, or label indicator array / sparse matrix
        Ground truth (correct) _labels.

    y_pred : 1d array-like, or label indicator array / sparse matrix
        Predicted _labels, as returned by a classifier.

    param_dict: dict
        A dictory saving the parameters including::

            sample_weight : array-like of shape = [n_samples], optional Sample
            weights.

    Returns
    -------
    score : float
    """
    # codes...

These functions always have three parameters: y_true for ground-truth labels, y_pred for prediction returned by an estimator, param_dict is a dict that stores other parameters used in the function.

The built-in metrics include:

'accuracy_score',
'zero_one_loss',
'roc_auc_score',
'get_fps_tps_thresholds',
'f1_score',
'hamming_loss',
'one_error',
'coverage_error',
'label_ranking_loss',
'label_ranking_average_precision_score',
'micro_auc_score',
'Average_precision_score',
'minus_mean_square_error'

Please refer to s3l.metrics.performance for more details.