I'm having trouble using the cov function for a pandas dataframe. Can someone explain why the first version is erroneously returning zeros?
In[2]: temp.cov()
Out[2]:
dogs cats
dogs 0.0 0.0
cats 0.0 0.0
In[3]: temp.iloc[1:,:].cov()
Out[3]:
dogs cats
dogs 271.813913 49.549741
cats 49.549741 11.500842
The dataframe is below...
dogs cats
0 -7.130125 -0.042310
1 -12.302285 -1.852217
2 -20.583190 -3.504634
3 -6.589786 -1.631955
4 19.512195 3.250522
5 18.612521 3.459353
6 7.029877 -1.525133
7 23.172906 3.555051
8 18.691589 1.913509
9 -27.184466 -7.424187
10 31.307551 6.743372
11 13.628620 -1.604586
12 8.756567 3.338248
13 -3.565062 -0.082786
14 3.539823 0.081160
15 -1.782531 -0.027595
16 -12.433393 -1.806514
{'dogs': {0: -7.1301247771835552, 1: -12.302284710017597, 2: -20.583190394511174, 3: -6.5897858319604596, 4: 19.51219512195118, 5: 18.612521150592254, 6: 7.0298769771528491, 7: 23.172905525846723, 8: 18.691588785046733, 9: -27.184466019417485, 10: 31.307550644567193, 11: 13.628620102214661, 12: 8.7565674255691839, 13: -3.5650623885917776, 14: 3.5398230088495297, 15: -1.7825311942958333, 16: -12.433392539964562}, 'cats': {0: -0.04231002139298834, 1: -1.8522167487684538, 2: -3.5046343844700178, 3: -1.6319546791088158, 4: 3.2505224053865356, 5: 3.4593526068693037, 6: -1.5251327504292034, 7: 3.5550505255468323, 8: 1.9135093761959987, 9: -7.424187066130683, 10: 6.7433720685594061, 11: -1.604586062335045, 12: 3.3382478256752668, 13: -0.082786254170352791, 14: 0.081160267179636314, 15: -0.027595418056747256, 16: -1.8065138668443836}}
also, if someone could show me an easy way to format data like this so that others could easily use it, I would appreciate it / will update it.
This appears to be a known issue with numpy and a recent version MKL. Running update conda --all seems to fix my issue.
It's well documented here... https://github.com/numpy/numpy/issues/9758