How to Pivot/Stack for multi header column dataframe

user_19720020 0
Viewed 38
np.random.seed(2022)  # added to make the data the same each time
cols = pd.MultiIndex.from_arrays([['A','A' ,'B','B'], ['min','max','min','max']])
df = pd.DataFrame(np.random.rand(3,4),columns=cols)
df.index.name = 'item'

             A                   B          
           min       max       min       max
item                                        
0     0.009359  0.499058  0.113384  0.049974
1     0.685408  0.486988  0.897657  0.647452
2     0.896963  0.721135  0.831353  0.827568

There are two column headers and while working with csv, I get a blank column name for every other column on unmerging.

I want result that looks like this. How can I do it? I tried to use pivot table but couldn't do it.

enter image description here

pandas pivot-table python
1 Answers

Try:

df = (
    df.stack(level=0)
    .reset_index()
    .rename(columns={"level_1": "title"})
    .sort_values(by=["title", "item"])
)
print(df)

Prints:

   item title       max       min
0     0     A  0.762221  0.737758
2     1     A  0.930523  0.275314
4     2     A  0.746246  0.123621
1     0     B  0.044137  0.264969
3     1     B  0.577637  0.699877
5     2     B  0.601034  0.706978

Then to CSV:

df.to_csv('out.csv', index=False)
user_100359850
Related