best way to use groupby or aggregate in pandas

Question

I have a table called clients and I want to show how many times someone registered or purchased an item based on the userid.

The goal is to have a table that shows the sum of registration_complete and purchased based on userid

This is the code that I wrote. Unfortunately not all columns show up

  new_file= new_data.groupby(['userid']) 
  ['Registration_Complete','Purchase'].agg('sum')
  new_file.head(5)

This is the table I'm using to figure out the registration and purchased based on userid

 Event_day  timestamp        install  userid  registration   purchase
 1/1/1900   1/1/1900 16:10    yes     555221     1               0
 1/1/1900   1/1/1900 16:12    yes     555221     1               1
 2/19/2010  1/19/2010 16:40   no      533211     0               1
 2/19/2010  1/19/2016 16:53   yes     533211     0               1
 2/20/2017  2/20/2017 15:46   yes     53200      1               0
 3/15/2017  3/15/2018 15:48   yes     53200      1               0
 3/15/2017  3/15/2018 20:14   yes     53200      1               0

and I want something that will give me the sum

Event_day  timestamp        install  userid  registration   purchase
1/1/1900   1/1/1900 16:10    yes     555221     2               0
2/19/2010  1/19/2016 16:53   yes     533211     0               2
3/15/2017  3/15/2018 20:14   yes     53200      5               0

Answer 1

IIUC you can keep the first or last value of other columns passing a dict to agg

agg = {'Event_day': 'last', 'timestamp': 'last', 'install': 'last', 'registration': 'sum', 'purchase': 'sum'}
df.groupby('userid').agg(agg).reset_index()

    userid  Event_day   timestamp       install registration    purchase
0   53200   3/15/2017   3/15/2018 20:14 yes     3               0
1   533211  2/19/2010   1/19/2016 16:53 yes     0               2
2   555221  1/1/1900    1/1/1900 16:12  yes     2               1

---

Edit:

Keeping in mind that several answers are possibly correct, I find it interesting to do a performance test between them

Timings

dfg1 = df.groupby("userid")["install", "timestamp", "Event_day"].max()
dfg2 = df.groupby("userid").sum()
pd.concat([dfg1, dfg2], axis=1)

38.5 ms ± 393 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)

first_other_columns = df[['Event_day', 'timestamp', 'install',  'userid']].drop_duplicates(['userid'], keep='first')
grouped = df.groupby(['userid']).sum().reset_index()
pd.merge(grouped, first_other_columns, on=['userid'])

11.3 ms ± 100 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)

agg = {'Event_day': 'last', 'timestamp': 'last', 'install': 'last', 'registration': 'sum', 'purchase': 'sum'}
df.groupby('userid').agg(agg).reset_index()

6.85 ms ± 62.5 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)