binning.py 文件源码

def _first_interval(x, n_bins):
    """
    Gets the first interval based on the percentiles, 
    either a closed interval containing the same value multiple times
    or a closed-open interval with a different lower and upper bound.
    """
    # calculate the percentiles
    percentiles = np.linspace(0., 100., n_bins + 1)
    bounds = np.percentile(x, q=percentiles, interpolation='higher')
    lower = bounds[0]
    upper = bounds[1]

    if lower == upper:
        return lower, upper, True, True
    else:
        return lower, upper, True, False


#------- private methods for categorical binnings-------#