根据多个条件创建新列和行

Question

I have the following dataframe:-

import pandas as pd
df = pd.read_csv('filename.csv')

print(df)

date        organic         paid        source_type
4/1/2018    39911909.19     38575924.75 Search
4/1/2018    5085939.952     882.608927  Social
4/1/2018    16227439.73     0           Mail
4/1/2018    0               5671871.24  Display Ads
4/1/2018    91215520.23     0           Direct
4/1/2018    15743479.56     0           Referrals

I want to add a column total_sum for all the source types except when source type is "Search". If the source_type is search I want to break down the single row into two and source type becomes organic search and paid search. Inshort a df like below. The summing part is easy to handle i am just stuck with the breaking of rows and conditional column prefix part. Dataframe I need:-

date        source_type     Total Sum
4/1/2018    Organic Search  39911909.19
4/1/2018    Paid Search     38575924.75
4/1/2018    Social          5086822.561
4/1/2018    Mail            16227439.73
4/1/2018    Display Ads     5671871.24
4/1/2018    Direct          91215520.23
4/1/2018    Referrals       15743479.56

Answer 1

You can split DataFrame by boolean indexing with Series.eq for == , then reshape first by DataFrame.melt with new column with Series.str.capitalize , filter second by invert mask by ~ , sum values with DataFrame.pop for remove column after and last use concat :

mask = df['source_type'].eq('Search')

df1 = df[mask].melt(['date','source_type'], value_name='Total Sum')
df1['source_type'] =  df1.pop('variable').str.capitalize() + ' Search'

df2 = df[~mask].copy()
df2['Total Sum'] = df2.pop('organic').add(df2.pop('paid'))
df = pd.concat([df1, df2], ignore_index=True)
print (df)
       date     source_type     Total Sum
0  4/1/2018  Organic Search  3.991191e+07
1  4/1/2018     Paid Search  3.857592e+07
2  4/1/2018          Social  5.086823e+06
3  4/1/2018            Mail  1.622744e+07
4  4/1/2018     Display Ads  5.671871e+06
5  4/1/2018          Direct  9.121552e+07
6  4/1/2018       Referrals  1.574348e+07