source_weights.py 文件源码

def normalize_source_weights(points, wcounts):
    wsum = 0.0
    wcounts_sum = 0
    for p in points:
        wsum += p.weight * wcounts[p.tag]
        wcounts_sum += wcounts[p.tag]

    print("The summation of window counts: %d" % wcounts_sum)
    print("The iniital summation(weight * window_counts): %f" % wsum)
    factor = 1.0 / wsum

    weights = {}
    for p in points:
        weights[p.tag] = p.weight * factor

    # validate
    wsum = 0.0
    for event in weights:
        wsum += wcounts[event] * weights[event]
    if not np.isclose(wsum, 1.0):
        raise ValueError("Error normalize source weights: %f" % wsum)
    print("The normalized sum is: %f" % wsum)
    print("Final weights: %s" % weights)
    return weights