CCDF.py

def plot_ccdf(data):
    """
    Plot the complementary cumulative distribution function
    (1-CDF(x)) based on the data on the axes object.
    Note that this way of computing and plotting the ccdf is not
    the best approach for a discrete variable, where many
    observations can have exactly same value!
    """
    # Note that, here we use the convention for presenting an
    # empirical 1-CDF (ccdf) as discussed
    # a quick way of computing a ccdf (valid for continuous data):
    sorted_vals = np.sort(np.unique(data))
    ccdf = np.zeros(max(sorted_vals))
    n = float(len(data))
    for i in range(1,max(sorted_vals)):
        ccdf[i] = np.sum(data >= i)/n
    return ccdf