Let's draw non-correlated values

The values are independently drawn from uniform random distributions



In [1]:

    
# ==  Basic import == #
# plot within the notebook
%matplotlib inline
# No annoying warnings
import warnings
warnings.filterwarnings('ignore')



In [32]:

    
import numpy as np



In [ ]:

    
x = np.random.rand(200)
y = np.random.rand(200)

See how it looks like



In [34]:

    
import matplotlib.pyplot as mpl
mpl.plot(x,y, ls="None",marker="s",ms=15, mfc="None")









    Out[34]:





[<matplotlib.lines.Line2D at 0x108844290>]

Now let's create a variable that depends on x "z" plus some other random distribution (noise or other dependency)



In [36]:

    
z = x*3 + np.random.normal(0,scale=0.3,size=200)



In [37]:

    
mpl.plot(x,z, ls="None",marker="s",ms=15, mfc="None")









    Out[37]:





[<matplotlib.lines.Line2D at 0x108937fd0>]

And measure the Pearson Correlation coefficient



In [38]:

    
from scipy import stats

First value rho, second p-value (see future lesson)



In [39]:

    
stats.pearsonr(x,y)









    Out[39]:





(-0.0421453339804998, 0.55348068796395489)



In [40]:

    
stats.pearsonr(x,z)









    Out[40]:





(0.94483025261099896, 6.3845709192655321e-98)



In [ ]: