Data Splitting, Melting, and Transformation in Python-CodePudding

I have the following code:

split_melt_and_transform_this_df = {'Concat' : [('AA,ABC,1,10,|','BB,ABC,,1,Z|','BB,ABC,,1,Z|''CC,ABC,50,,Z|'),('AA,ABC,1,10,|','BB,ABC,1,,Z|'),('AA,ABC,1,10,|','BB,ABC,1,,Z|'),('AA,ABC,5,,z|')], 'Item' : ['Backstreet Boys','NYSYNC', 'One Direction', 'Nirvana']}
split_melt_and_transform_this_df = pd.DataFrame(split_this_df)
split_melt_and_transform_this_df

I'd like to transform this 'Current DF' into the following 'Desired DF1' and 'Desired DF2' data frames:

DF2 is the especially tricky part. All 'Z' values that end up in the final column need to be infinity. If the row in the final column is blank, then show the value in the row of the adjacent column (i.e., second to last column, where the '10' value shows).

CodePudding user response：

You can try explode to get df1

df1 = df.explode('Concat', ignore_index=True)

print(df1)

          Concat             Item
0  AA,ABC,1,10,|  Backstreet Boys
1   BB,ABC,,1,Z|  Backstreet Boys
2   BB,ABC,,1,Z|  Backstreet Boys
3  CC,ABC,50,,Z|  Backstreet Boys
4  AA,ABC,1,10,|           NYSYNC
5   BB,ABC,1,,Z|           NYSYNC
6  AA,ABC,1,10,|    One Direction
7   BB,ABC,1,,Z|    One Direction
8   AA,ABC,5,,Z|          Nirvana

To get df2, you can try to split the Concat column and convert list column to dataframe.

df2 = df1.join(pd.DataFrame(df1.pop('Concat').str.split(',').tolist(), index=df1.index))
df2[4] = np.select([df2[4].eq('|'), df2[4].eq('Z|')], [df2[3], np.inf])

print(df2)

              Item   0    1   2   3    4
0  Backstreet Boys  AA  ABC   1  10   10
1  Backstreet Boys  BB  ABC       1  inf
2  Backstreet Boys  BB  ABC       1  inf
3  Backstreet Boys  CC  ABC  50      inf
4           NYSYNC  AA  ABC   1  10   10
5           NYSYNC  BB  ABC   1      inf
6    One Direction  AA  ABC   1  10   10
7    One Direction  BB  ABC   1      inf
8          Nirvana  AA  ABC   5      inf