pandas pivot_table column names

Question

For a dataframe like this:

d = {'id': [1,1,1,2,2], 'Month':[1,2,3,1,3],'Value':[12,23,15,45,34], 'Cost':[124,214,1234,1324,234]}
df = pd.DataFrame(d)

     Cost  Month  Value  id  
0    124       1     12   1  
1    214       2     23   1  
2    1234      3     15   1  
3    1324      1     45   2  
4    234       3     34   2  

to which I apply pivot_table

df2 =    pd.pivot_table(df, 
                        values=['Value','Cost'],
                        index=['id'],
                        columns=['Month'],
                        aggfunc=np.sum,
                        fill_value=0)

to get df2:

       Cost            Value          
Month     1    2     3     1   2   3   
id                                  
1       124  214  1234    12  23  15
2      1324    0   234    45   0  34

is there an easy way to format resulting dataframe column names like

id     Cost1    Cost2     Cost3 Value1   Value2   Value3   
1       124      214      1234    12        23       15
2      1324       0       234     45         0       34

If I do:

df2.columns =[s1 + str(s2) for (s1,s2) in df2.columns.tolist()]

I get:

    Cost1  Cost2  Cost3  Value1  Value2  Value3
id                                             
1     124    214   1234      12      23      15
2    1324      0    234      45       0      34

How to get rid of the extra level?

thanks!

Mulcahy

Answer 1

Using clues from @chrisb's answer, this gave me exactly what I was after:

df2.reset_index(inplace=True)

which gives:

id     Cost1    Cost2     Cost3 Value1   Value2   Value3   
1       124      214      1234    12        23       15
2      1324       0       234     45         0       34

and in case of multiple index columns, this post explains it well. just to be complete, here is how:

df2.columns = [' '.join(col).strip() for col in df2.columns.values]

Mulcahy

"Flatten Hierarchical Index" from Community's post helped me with the same issue. df.columns = [' '.join(col).strip() for col in df.columns.values] – Cleodel

Answer 2

'id' is the index name, which you can set to None to remove.

In [35]: df2.index.name = None

In [36]: df2
Out[36]: 
   Cost1  Cost2  Cost3  Value1  Value2  Value3
1    124    214   1234      12      23      15
2   1324      0    234      45       0      34

Until

Answer 3

I don't think reset_index is the correct way to do this. If df.columns.name is what you try to remove, perhaps you should try:

df.columns.name = None

Slick