CODEWITHSUNDEEP
pythonpandas
danuker
danuker

Reputation: 881

In Pandas, creating a Panel of MultiIndex'ed DataFrames

I am creating a Panel out of two DataFrames as such:

import pandas as pd

d1 = pd.DataFrame(index=pd.MultiIndex.from_product(iterables=[['a','b'],[1,'2','3'],[4]]), data=list(range(6)))
print('d1 before panel:')
print(d1)
d2 = pd.DataFrame(index=pd.MultiIndex.from_product(iterables=[['a'],['1','2'],[1, '2']]), data=list(range(4)))
print('\nd2 before panel:')
print(d2)

print('=================')

p = pd.Panel({'d1': d1, 'd2': d2})
print('\npanel d1:')
print(p['d1'])
print('\npanel d2:')
print(p['d2'])

Note that I mixed index types (ints and strings).

The Panel I get after this initialization shows d1 completely NaN, with some indices duplicated ('a'-1-4 and 'a'-'2'-4).

It seems that something has gone horribly wrong. Are there any documented no-no's that I'm breaching?

Upvotes: 1

Views: 514

Answers (1)

danuker
danuker

Reputation: 881

I solved this problem by updating Pandas. I was using 0.13, and now I have 0.15.2.

Upvotes: 1

Related Questions