metrics.py 文件源码

def f1_same_duplicates_score(x, y):
    """
    Given cluster labels x and y, compute the f1 score
    that the same elements are marked as duplicates
    """
    import warnings
    from sklearn.metrics import f1_score

    if x.shape != y.shape:
        raise ValueError
    x_dup = _dbscan_unique2noisy(x)
    x_dup[x_dup > -1] = 1  # duplicates
    x_dup[x_dup == -1] = 0  # not duplicates
    y_dup = _dbscan_unique2noisy(y)
    y_dup[y_dup > -1] = 1  # duplicates
    y_dup[y_dup == -1] = 0  # not duplicates

    x_dup = np.abs(x_dup)
    y_dup = np.abs(y_dup)

    with warnings.catch_warnings():
        warnings.filterwarnings("ignore", category=UndefinedMetricWarning)
        score = f1_score(x_dup, y_dup)

    return score