Behavior of Pandas DataFrame .ix method

Question

Here are 2 data frames that I would like to be concat together column wise.

>>> df1 = pd.DataFrame({'letters' : ['a', 'b', 'c'], 'numbers' : [1, 2, 3]})
>>> df2 = pd.DataFrame({'Cities' : ['Rome', 'Venice'], 'floats' : [1.1 , 2.2]})

>>> df1
  letters  numbers
0       a        1
1       b        2
2       c        3

>>> df2
   Cities  floats
0    Rome     1.1
1  Venice     2.2

There is a mismatch in the number of rows. I would like to append a copy of the second row (could be any arbitrary row) so I tried this...

>>> df2.ix[[0, 1, 1]]
   Cities  floats
0    Rome     1.1
1  Venice     2.2
1  Venice     2.2

When concating the 2 data frames I get a ValueError...

pd.concat([df1, df2.ix[[0, 1, 1]]], axis = 1)
ValueError: Shape of passed values is (4, 6), indices imply (4, 4)

I tried making a new copy of the table with the replicated row with no avail...

pd.concat([df1, df2.ix[[0, 1, 1]]].copy(), axis = 1)
ValueError: Shape of passed values is (4, 6), indices imply (4, 4)

This is more of a contrived example for understanding than an actual problem as the premise is a little silly. I would still like a proper answer though with an expected result of...

  letters  numbers  Cities  floats
0       a        1    Rome     1.1
1       b        2  Venice     2.2
2       c        3  Venice     2.2

Answer 1

pd.concat aligns rows based on the index of the DataFrames. Since df2.ix[...] has two rows with the same index, pd.concat does not place the second "Venice" row on a row with index 2. To renumber the index, call reset_index() before concat'ing:

In [102]: pd.concat([df1, df2.iloc[[0, 1, 1]].reset_index()], axis=1)
Out[102]: 
  letters  numbers  index  Cities  floats
0       a        1      0    Rome     1.1
1       b        2      1  Venice     2.2
2       c        3      1  Venice     2.2