asit-pal

mean = np.array([0,0])
rho=0.8
cov = np.array([[1,rho],[rho,1]])
distr = multivariate_normal(mean=mean,cov=cov,size=5000)
X,Y = distr[:,0],distr[:,1]

asit-pal

from sklearn.feature_selection import mutual_info_regression as MIR
mutual_information = []

for i in range(1000,100000,1000):
    distr = multivariate_normal(mean=mean,cov=cov,size=i)
    X,Y = distr[:,0],distr[:,1]
    mutual_information.append(MIR(X.reshape(-1,1),Y))

asit-pal

plt.axhline(ana,lw=2,color='green',label='Analytical solution')
plt.plot(range(1000,100000,1000),mutual_information,label='With scikit-learn function')
plt.xlabel('No of samples')
plt.ylabel('Mutual information')
plt.legend()
plt.savefig('Mutual_information_convergence_test.png',dpi=300)