[FIXED] Why I get NAN when calculate corr of those two series?

Issue

I have two pandas Series object:

print(s1)
print("="*80)
print(s2)

# output

0   -0.443538
1   -0.255012
2   -0.582948
3   -0.393485
4    0.430831
5    0.232216
6   -0.014269
7   -0.133158
8    0.127162
9   -1.855860
Name: s1, dtype: float64
================================================================================
29160   -0.650857
29161   -0.135428
29162    0.039544
29163    0.241506
29164   -0.793352
29165   -0.054500
29166    0.901152
29167   -0.660474
29168    0.098551
29169    0.822022
Name: s2, dtype: float64

And I want to calculate corr of those two series:

s1.corr(s2)

#output

nan

I don't know why I get 'nan' here, using numpy gives the correct result:

np.corrcoef(s1,s2)[0][1]

#output

-0.4918385039519204

Did I do something wrong in the above code?

Solution

Your series indices are not aligned. Pandas realigns s2 on s1 yielding only NaNs.

You can force the index of s1 on s2 using set_axis:

s1.corr(s2.set_axis(s1.index))

Output: -0.49183852303556697

Answered By - mozway