Flattening MultiIndex in header swaps the column names

Question

I have the following data frame with multiple headers:

    Datetime            Value
id                      a           b           c           d           e
0   2017-01-01 00:00:00 0.774016    1.588788    270.06055   268.9109    93060.31
1   2017-01-01 00:10:00 0.774016    1.588788    270.06055   268.9109    93060.31
2   2017-01-01 00:20:00 0.774016    1.588788    270.06055   268.9109    93060.31
3   2017-01-01 00:30:00 0.774016    1.588788    270.06055   268.9109    93060.31
4   2017-01-01 00:40:00 0.774016    1.588788    270.06055   268.9109    93060.31

When I pass from multiple headers to a single header, at some point the column names are swapped and I don't know how to fix it.

cols = ["a","b","c","d","e"]
df.columns = [col[1] if col[0] == '' else col[0] for col in df.columns]
cols.insert(0,"Datetime")
df.columns = cols

This gives me swapped column names:

    Datetime            a           b           d           e           c
0   2017-01-01 00:00:00 0.774016    1.588788    270.06055   268.9109    93060.31
1   2017-01-01 00:10:00 0.774016    1.588788    270.06055   268.9109    93060.31
2   2017-01-01 00:20:00 0.774016    1.588788    270.06055   268.9109    93060.31
3   2017-01-01 00:30:00 0.774016    1.588788    270.06055   268.9109    93060.31
4   2017-01-01 00:40:00 0.774016    1.588788    270.06055   268.9109    93060.31

How can I fix it?

Update:

{('Datetime', ''): {0: Timestamp('2017-01-01 00:00:00'),
  1: Timestamp('2017-01-01 00:10:00'),
  2: Timestamp('2017-01-01 00:20:00'),
  3: Timestamp('2017-01-01 00:30:00'),
  4: Timestamp('2017-01-01 00:40:00')},
 ('Value', 'a'): {0: 0.774016,
  1: 0.774016,
  2: 0.774016,
  3: 0.774016,
  4: 0.774016},
 ('Value', 'b'): {0: 1.588788,
  1: 1.588788,
  2: 1.588788,
  3: 1.588788,
  4: 1.588788},
 ('Value', 'c'): {0: 270.06055,
  1: 270.06055,
  2: 270.06055,
  3: 270.06055,
  4: 270.06055},
 ('Value', 'd'): {0: 268.9109,
  1: 268.9109,
  2: 268.9109,
  3: 268.9109,
  4: 268.9109},
 ('Value', 'e'): {0: 93060.31,
  1: 93060.31,
  2: 93060.31,
  3: 93060.31,
  4: 93060.31}}

Answer 1

Bruteforce approach

>>> pd.concat([df[['Datetime']].droplevel(1, axis=1), df["Value"]], axis=1)
               Datetime         a         b          c         d         e
id
0   2017-01-01 00:00:00  0.774016  1.588788  270.06055  268.9109  93060.31
1   2017-01-01 00:10:00  0.774016  1.588788  270.06055  268.9109  93060.31
2   2017-01-01 00:20:00  0.774016  1.588788  270.06055  268.9109  93060.31
3   2017-01-01 00:30:00  0.774016  1.588788  270.06055  268.9109  93060.31
4   2017-01-01 00:40:00  0.774016  1.588788  270.06055  268.9109  93060.31

Answer 2

Try with set_index + droplevel + reset_index :

df.set_index('Datetime', append=True).droplevel(0, 1).reset_index('Datetime')

              Datetime         a         b          c         d         e
id                                                                       
0  2017-01-01 00:00:00  0.774016  1.588788  270.06055  268.9109  93060.31
1  2017-01-01 00:10:00  0.774016  1.588788  270.06055  268.9109  93060.31
2  2017-01-01 00:20:00  0.774016  1.588788  270.06055  268.9109  93060.31
3  2017-01-01 00:30:00  0.774016  1.588788  270.06055  268.9109  93060.31
4  2017-01-01 00:40:00  0.774016  1.588788  270.06055  268.9109  93060.31

---

Also to fix your implementation, don't insert into the list and misalign the DataFrame just do:

df.columns = [col[1] if col[1] else col[0] for col in df.columns]

              Datetime         a         b          c         d         e
id                                                                       
0  2017-01-01 00:00:00  0.774016  1.588788  270.06055  268.9109  93060.31
1  2017-01-01 00:10:00  0.774016  1.588788  270.06055  268.9109  93060.31
2  2017-01-01 00:20:00  0.774016  1.588788  270.06055  268.9109  93060.31
3  2017-01-01 00:30:00  0.774016  1.588788  270.06055  268.9109  93060.31
4  2017-01-01 00:40:00  0.774016  1.588788  270.06055  268.9109  93060.31