Python根据条件将值转移到列

Question

I am facing a problem related to a delivery offset in the example dataset.

df = pd.DataFrame({'Platform ID' : [1,2,3,4], "2018" : [0,0,0,0],
                                              "2019" : [0,0,0,0], 
                                              "2020" : [21,3,8,23], 
                                              "2021" : [31,0,10,12], 
                                              "2022" : [2,23,13,14],
                   'Offsets (in months)' : [6, 12, 18, 0]})
df

What I need is a time shift based on the last column if the delivery is 6 months the half of the value should go to the previous year's column. If the time shift is 12 months the whole value must move to the previous year's column. If the time shift is 18 the value must move to the previous column and one-half of it must move even further like by 6 months in the first case.

I tried to use shift() however it moves the whole column. Is there a way how to set this condition in this or another function?

Desired output:

df = pd.DataFrame({'Platform ID' : [1,2,3,4], "2018" : [0,0,4,0],
                                              "2019" : [10.5,3,9,0], 
                                              "2020" : [26,0,11.5,23], 
                                              "2021" : [16.5,23,6.5,12], 
                                              "2022" : [1,0,0,14],
                   'Offsets (in months)' : [6, 12, 18, 0]})
df

Answer 1

This is in no way the most elegant solution out there, but given that each row essentially follows a different functional shift, I can't think of a much neater way of doing it than the below. I have amended my previous answer to adjust it to the new provided information. If a neater answer exists, I would be curious to see it.

df = pd.DataFrame({'Platform ID' : [1,2,3,4], "2018" : [0,0,0,0],
                                              "2019" : [0,0,0,0], 
                                              "2020" : [21,3,8,23], 
                                              "2021" : [31,0,10,12], 
                                              "2022" : [2,23,13,14],
                   'Offsets (in months)' : [6, 12, 18, 0]})
for i in df['Offsets (in months)'].unique():
    if i == 6:
        df_6 = df[df['Offsets (in months)'] == 6]
        df_6.iloc[:,1:6] = (df_6.iloc[:,1:6]/2).shift(periods=-1, fill_value=0, axis=1) +df_6.iloc[:,1:6]/2
        df[df['Offsets (in months)'] == 6] = df_6
    elif i == 12:
        df_12 = df[df['Offsets (in months)'] == 12]
        df_12.iloc[:,1:6] = df_12.iloc[:,1:6].shift(periods=-1, fill_value=0, axis=1)
        df[df['Offsets (in months)'] == 12] = df_12
    elif i == 18:
        df_18 = df[df['Offsets (in months)'] == 18]
        df_18.iloc[:,1:6] = df_18.iloc[:,1:6].shift(periods=-1, fill_value=0, axis=1)
        df_18.iloc[:,1:6] = (df_18.iloc[:,1:6]/2).shift(periods=-1, fill_value=0, axis=1) +df_18.iloc[:,1:6]/2
        df[df['Offsets (in months)'] == 18] = df_18

Unless there is a typo, this should do the trick. I am rushed so I might have made a mistake. Basically, the idea is to filter out each of the offsets using boolean indexing and then shifting the values excluding Platform ID and Offset (by specifying iloc[1:6].