Index._join_level on non-unique index is not implemented

Question

Here's my data

my dataframe name is

tab

        user_id     len

        barcode

0        1          490

1        2          71

2        3          1

3        5          1

This is my code

tab = tab[tab[('len', 'barcode')]]

tab.columns

MultiIndex(levels=[['len', 'user_id'], ['barcode', '']],

           labels=[[1, 0], [1, 0]])

This code on my routine notebook and works more than 10 times, but it not works anymore

Answer 1

I think there is problem your second level name is empty string.

So need:

a = tab[('len', '')]
print (a)
0    490
1     71
2      1
3      1
Name: (len, ), dtype: int64

If you want replace all empty strings in columns names:

tab = tab.rename(columns={'':'b'})
a = tab[('len', 'b')]
print (a)
0    490
1     71
2      1
3      1
Name: (len, b), dtype: int64

But better is remove [] for remove MultiIndex in columns:

tab = pd.pivot_table(barcode,index="user_id",values="barcode",‌aggfunc='size')
tab = tab.reset_index()

Sign up