Python pandas pivot vs pivot_table

Question

If I have a multiindex DataFrame in Pandas, if I read through the documentation of pivot and pivot_table I cannot seem to find the reason why pivot doesn't work in this example. Clearly I am missing something, but it takes the same parameters and seems to suggest that it would work. What am I missing here? Why does pivot throw an error but pivot_table works perfectly. Thanks.

# standard imports
import pandas as pd

# create a random multiindex dataframe
outside = ['G1','G1','G1','G2','G2','G2']
inside = [1,2,3,1,2,3]
hier_index = list(zip(outside,inside))
hier_index = pd.MultiIndex.from_tuples(hier_index)
df = pd.DataFrame(np.random.randn(6,1),index=hier_index,columns=['A'])
df.index.names = ['Group', 'Num']

# show the original dataframe
print(df)

# successfully pivots the dataframe as requested
print(df.pivot_table(index='Group', columns='Num', values='A'))

# causes a KeyError on 'Group' - not understanding why
print(df.pivot(index='Group', columns='Num', values='A'))

Answer 1

Here function pivot not working with MultiIndex , solution is DataFrame.reset_index for convert levels to columns:

print(df.reset_index().pivot(index='Group', columns='Num', values='A'))
Num           1         2         3
Group                              
G1    -0.121167 -2.008585 -0.920647
G2     0.168234 -1.319892  1.266429

I think reason why working pivot_table nice is hidden in groupby-enhancements - you can use groupby by levels and similar pivot_table .