from scipy import stats import matplotlib.pyplot as plt np.random.seed(1234) # make this example reproducible # Generate some data and determine optimal ``lmbda`` in various ways: x = stats.loggamma.rvs(5, size=30) + 5 y, lmax_mle = stats.boxcox(x) lmax_pearsonr = stats.boxcox_normmax(x) lmax_mle # 7.177... lmax_pearsonr # 7.916... stats.boxcox_normmax(x, method='all') # array([ 7.91667384, 7.17718692]) fig = plt.figure() ax = fig.add_subplot(111) prob = stats.boxcox_normplot(x, -10, 10, plot=ax) ax.axvline(lmax_mle, color='r') ax.axvline(lmax_pearsonr, color='g', ls='--') plt.show()