Build in function for computing covariance

Question

Is there a way in python to obtain the covariance matrix given the mean and sample data points

Example:

mean = [3 3.6]
data = [[1 2]
        [2 3]
        [3 3]
        [4 5] 
        [5 5]]

I know how to calculate the same by substituting these values in the formula. But is there a build in function in python which does this for me. I know there is one in Matlab, but I am not sure about python.

Answer 1

numpy.cov() can be used to compute the covariance matrix:

In [1]: import numpy as np

In [2]: data = np.array([[1,2], [2,3], [3,3], [4,5], [5,5]])

In [3]: np.cov(data.T)
Out[3]: 
array([[ 2.5,  2. ],
       [ 2. ,  1.8]])

By default, np.cov() expects each row to represent a variable, with observations in the columns. I therefore had to transpose your matrix (by using .T).

An alternative way to achieve the same thing is by setting rowvar to False:

In [15]: np.cov(data, rowvar=False)
Out[15]: 
array([[ 2.5,  2. ],
       [ 2. ,  1.8]])