In [1]:

    

import numpy as np
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
from scipy.stats import gamma
from scipy.special import gamma as g
from scipy.special import gammaincc
from math import factorial, exp
from itertools import permutations, combinations
from sklearn.metrics.pairwise import rbf_kernel, laplacian_kernel

%matplotlib inline


    

In [2]:

    

import sys
sys.path.insert(0, '/Users/mati/Devel/dsga1005/code')

from independence_test import *
from r_independence import *


    

In [3]:

    

size = 2000
X_1 = np.random.rand(size)
Y_1 = np.random.randn(size)


    

In [4]:

    

test_ind = HSIC_b(X_1, Y_1, kernel='exponential')


    

In [5]:

    

test_ind.empirical_test()


    

    
Out[5]:





5.394588047922858e-06

In [6]:

    

test_ind.p_value


    

    
Out[6]:





0.0

In [7]:

    

test_ind_r = dHSIC(X_1, Y_1)


    

In [28]:

    

print test_ind_r.res


    

    




$statistic
[1] 0.2921431

$crit.value
[1] 0.6047159

$p.value
[1] 0.4910761

$time
GramMat   dHSIC CritVal 
  1.219   0.037   0.053 

In [8]:

    

test_ind_r.statistic


    

    
Out[8]:





0.2921430944802239

In [9]:

    

test_ind_r.p_value


    

    
Out[9]:





0.49107605699028706

In [18]:

    

size = 200
X_2 = np.random.normal(0,10,size)
Y_2 = X_2


    

In [19]:

    

test_non = HSIC_b(X_2, Y_2, kernel='exponential')


    

In [20]:

    

test_non.empirical_test()


    

    
Out[20]:





0.025254149673630218

In [21]:

    

test_non.p_value


    

    
Out[21]:





1.0

In [22]:

    

test_non_r = dHSIC(X_2, Y_2)


    

In [29]:

    

print test_non_r.res


    

    




$statistic
[1] 20.15576

$crit.value
[1] 0.5838765

$p.value
[1] 1.843958e-132

$time
GramMat   dHSIC CritVal 
  0.009   0.001   0.001 

In [23]:

    

test_non_r.statistic


    

    
Out[23]:





20.155764972379085

In [24]:

    

test_non_r.p_value


    

    
Out[24]:





1.8439575315357636e-132

In [3]:

    

SIZE = 500
Z = np.random.randn(SIZE) * 100
X = Z + np.random.randn(SIZE)
Y = Z + np.random.randn(SIZE)

Z_vars = np.array(['Z'])

data = pd.DataFrame(np.array([X, Y, Z]).T, columns=["X", "Y", "Z"])


    

In [4]:

    

ci = CI('X', 'Y', ['Z'], data, 'cor')


    

In [6]:

    

ci.statistic


    

    
Out[6]:





0.04249053645171306

In [7]:

    

ci.p_value


    

    
Out[7]:





0.3435296910152168

In [8]:

    

SIZE = 500
Z = np.random.randn(SIZE) * 100
X = np.random.randn(SIZE)
Y = X

Z_vars = np.array(['Z'])

data = pd.DataFrame(np.array([X, Y, Z]).T, columns=["X", "Y", "Z"])


    

In [9]:

    

ci = CI('X', 'Y', ['Z'], data, 'cor')


    

In [11]:

    

ci.statistic


    

    
Out[11]:





-1.0

In [12]:

    

ci.p_value


    

    
Out[12]:





0.0

In [ ]: