pp_plot.py

def pp_plot(x, dist, line=True, ax=None):
    '''
    Function for comparing empirical data to a theoretical distribution by using a P-P plot.
    
    Params:
    x - empirical data
    dist - distribution object from scipy.stats; for example scipy.stats.norm(0, 1)
    line - boolean; specify if the reference line (y=x) should be drawn on the plot
    ax - specified ax for subplots, None is standalone
    '''
    if ax is None:
        ax = plt.figure().add_subplot(1, 1, 1)
        
    n = len(x)
    p = np.arange(1, n + 1) / n - 0.5 / n
    pp = np.sort(dist.cdf(x))
    sns.scatterplot(x=p, y=pp, color='blue', edgecolor='blue', ax=ax)
    ax.set_title('PP-plot')
    ax.set_xlabel('Theoretical Probabilities')
    ax.set_ylabel('Sample Probabilities')
    ax.margins(x=0, y=0)
    
    if line: plt.plot(np.linspace(0, 1), np.linspace(0, 1), 'r', lw=2)
    
    return ax