python matplotlib - contour plot - confidence intervals

matplotlib, python, statistics

Solution

You can specify a list of `z-values` where the contours are drawn. So all you have to do is collect the correct `z-values` for your distribution. Here is an example for '1, 2, and 3 sigma away from the peak value':

Code:

import numpy as np
import matplotlib.cm as cm
import matplotlib.mlab as mlab
import matplotlib.pyplot as plt

#Set up the 2D Gaussian:
delta = 0.025
x = np.arange(-3.0, 3.0, delta)
y = np.arange(-3.0, 3.0, delta)
X, Y = np.meshgrid(x, y)
sigma = 1.0
Z = mlab.bivariate_normal(X, Y, sigma, sigma, 0.0, 0.0)
#Get Z values for contours 1, 2, and 3 sigma away from peak:
z1 = mlab.bivariate_normal(0, 1 * sigma, sigma, sigma, 0.0, 0.0)
z2 = mlab.bivariate_normal(0, 2 * sigma, sigma, sigma, 0.0, 0.0)
z3 = mlab.bivariate_normal(0, 3 * sigma, sigma, sigma, 0.0, 0.0)

plt.figure()
#plot Gaussian:
im = plt.imshow(Z, interpolation='bilinear', origin='lower',
                 extent=(-50,50,-50,50),cmap=cm.gray)
#Plot contours at whatever z values we want:
CS = plt.contour(Z, [z1, z2, z3], origin='lower', extent=(-50,50,-50,50),colors='red')
plt.savefig('fig.png')
plt.show()

Problem

I'm trying to plot contours (doable) on a grid of data using matplotlib.pyplot.contour, but with the contours placed at 1, 2 and 3 sigma away from the peak value. Is there a neat way to do this apart from brute force? Thanks! Python version is Python 2.7.2 |EPD 7.2-2 (64-bit)| (default, Sep 7 2011, 16:31:15) [GCC 4.0.1 (Apple Inc. build 5493)] on darwin

Original source