熊猫将重复的列转换为行

Question

I have a dataframe like this with repeating column names: ID is loaded as index

          JANUARY         FEBRUARY        MARCH 
  ID    Sales   Revenue Sales   Revenue Sales   Revenue
  03    10.00   5.00    0.00    0.00    10.00   19.00
  05    20.00   20.00   20.00   20.00   20.00   20.00
  06    30.00   30.00   30.00   30.00   30.00   30.00
  07    30.00   30.00   30.00   30.00   30.00   30.00

I want to show it as below:

  ID    Sales   Revenue
  03    10.00   5.00
  05    20.00   20.00
  06    30.00   30.00
  07    30.00   30.00
  03    0.00    0.00
  05    20.00   20.00
  06    30.00   30.00
  07    30.00   30.00
  03    10.00   19.00
  05    20.00   20.00
  06    30.00   30.00
  07    30.00   30.00

Currently I'm using, but expecting a better way. I have tried melt, but that's only for one column:

cols = df.columns.to_list()
for i in range(1, len(cols), 2):  # #Loading each month's data to the data frame
    sub_cols = cols[i:i + 2]
    sub_cols .insert(0, cols[0])
    sub_df = df.filter(sub_cols , axis=1)
    sub_df.columns = ['ID', 'Revenue', 'Sales']
    if i == 1:
        final_df = sub_df
    else:
        final_df = final_df.append(sub_df)

Answer 1

Here's another way to stack the columns. Not sure if it's more efficient but it takes less code.

#        JANUARY         FEBRUARY        MARCH 
#  ID    Sales   Revenue Sales   Revenue Sales   Revenue
#  03    10.00   5.00    0.00    0.00    10.00   19.00
#  05    20.00   20.00   20.00   20.00   20.00   20.00
#  06    30.00   30.00   30.00   30.00   30.00   30.00
#  07    30.00   30.00   30.00   30.00   30.00   30.00

import pandas as pd
dd = {
'ID':['03','05','06','07'],
'Sales1':[10,20,30,30],
'Rev1':[5,20,30,30],
'Sales2':[0,20,30,30],
'Rev2':[0,20,30,30],
'Sales3':[10,20,30,30],
'Rev3':[19,20,30,30]
}

df = pd.DataFrame(dd)
print(df.to_string(index=False),'\n') # source dataframe

####################

dfnew = pd.DataFrame(columns = ['ID', 'Sales', 'Revenue'])  # new dataframe with all data
for c in range(1,len(df.columns),2):
   dftmp = df[['ID',df.columns[c],df.columns[c+1]]] # create df for each month
   dftmp.columns = ['ID', 'Sales', 'Revenue'] # must rename columns for append
   dfnew = dfnew.append(dftmp)  # append to stacked df

print(dfnew.to_string(index=False))

Output

 ID  Sales1  Rev1  Sales2  Rev2  Sales3  Rev3
 03      10     5       0     0      10    19
 05      20    20      20    20      20    20
 06      30    30      30    30      30    30
 07      30    30      30    30      30    30

 ID Sales Revenue
 03    10       5
 05    20      20
 06    30      30
 07    30      30
 03     0       0
 05    20      20
 06    30      30
 07    30      30
 03    10      19
 05    20      20
 06    30      30
 07    30      30

Answer 2

Pandas lreshape did the trick for me.

df = pd.lreshape(df, 
{'Sales': file_df.columns[file_df.columns.str.match(r'^Sales\.?\d?')],
'Revenue': file_df.columns[file_df.columns.str.match(r'^Revenue\.?\d?')]})