How can I create column headers from row values in a slice of columns resulting in the column row values matching the column headers?

Question

I am wanting to convert dataframe row values into column headers, but I would only want one column of that same name. Then I want the row values under each new column to be the same as the column header. I have tried looking at get_dummies, pivot, pivot_table, group_by. Any help would be much appreciated.

Data:

import pandas as pd

In [1]: df = pd.DataFrame([['foo', 'bar', 'ham'], ['baz', 'foo', 'bar'], ['ham', 'baz', 'egg']], columns=['A', 'B', 'C'])


In [2]: df
Out[2]: 
     A    B    C
0  foo  bar  ham
1  baz  foo  bar
2  ham  baz  egg

I want this:

In [3]: 
Out[3]: 
   foo  bar  baz  ham  egg
0  foo  bar       ham
1  foo  bar  baz  
2            baz  ham  egg

Thank you.

Answer 1

The idea is to convert the DataFrame to the long form and then pivot:

df = df.reset_index()
df = pd.melt(df, id_vars=['index']).pivot(index = 'index', values='variable', columns='value')
for c in df.columns:
    df.loc[df[c].notna(), c] = c

value   bar baz egg foo ham
index                   
0       bar NaN NaN foo ham
1       bar baz NaN foo NaN
2       NaN baz egg NaN ham

Detailed explanation:

melt would convert the DataFrame to the following form:

   index    variable    value
0   0   A   foo
1   1   A   baz
2   2   A   ham
3   0   B   bar
4   1   B   foo
5   2   B   baz
6   0   C   ham
7   1   C   bar
8   2   C   egg

Then use pivot to make the columns to be all the unique values:

value   bar baz egg foo ham
index                   
0   B   NaN NaN A   C
1   C   A   NaN B   NaN
2   NaN B   C   NaN A

Then simply replace all non-na columns to be the column name.