98d8f98800a76e6809c5ac43245bfe685a5cf436,mir_eval/structure.py,,pairwise,#Any#Any#Any#Any#Any#Any#,69

Before Change

    matches     = 0.0
    n_agree_ref = 0.0
    n_agree_est = 0.0
    for i in xrange(len(y_ref)):
        for j in xrange(i + 1, len(y_ref)):
            // Do i and j match in reference?
            ref_match = (y_ref[i] == y_ref[j])
            n_agree_ref += ref_match

            // Or in estimate?
            est_match = (y_est[i] == y_est[j])
            n_agree_est += est_match

            // If both, we have agreement
            matches += (ref_match & est_match)
            
    precision   = matches / n_agree_est
    recall      = matches / n_agree_ref
    f_measure   = util.f_measure(precision, recall, beta=beta)

After Change

    n_agree_ref = (agree_ref.sum() - len(y_ref)) / 2.0

    // Repeat for estimate
    agree_est   = np.equal.outer(y_est, y_est)
    n_agree_est = (agree_est.sum() - len(y_est)) / 2.0

    // Find where they agree
    matches     = np.logical_and(agree_ref, agree_est)
    n_matches   = (matches.sum() - len(y_ref)) / 2.0

Instances