[FIXED] Filter out groups in pandas based on values in groups

Issue

Currently i have a dataframe looking something like this

Session Id	Time	Event	Data
1	10:00	Btn click	foo
1	11:00	Identification	bar
2	...	Btn click	foo
2	...	Btn click	foo
3	..	Identification	bar

I want to group my data by Session Id and process the groups further, but only if they possess an Identification Event.

Currently my solution looks like this:

for session in df.groupby('Session ID'):
    session_df: pd.DataFrame = session[1]
    if 'Identification' in session_df['Event'].values:
        process(session_df)

I tried to use filter on the groupby but got something wrong:

for session in df.groupby('Session ID').filter(lambda s: 'Identification' in s['Event'].values):
    process(session[1])

Solution

Code

make codition by groupby + transform

cond = df['Event'].eq('Identification').groupby(df['Session Id']).transform(sum).gt(0)
out = df[cond]

out :

    Session Id  Time    Event           Data
0   1          10:00    Btn click       foo
1   1          11:00    Identification  bar
4   3          ..       Identification  bar

if you want groupby + filter, use following code:

df.groupby('Session Id').filter(lambda x: x['Event'].eq('Identification').sum() > 0)

same result

---

Example Code

import pandas as pd
data1 = {'Session Id': [1, 1, 2, 2, 3], 
         'Time': ['10:00', '11:00', '...', '...', '..'], 
         'Event': ['Btn click', 'Identification', 'Btn click', 'Btn click', 'Identification'], 
         'Data': ['foo', 'bar', 'foo', 'foo', 'bar']}
df = pd.DataFrame(data1)

Answered By - Panda Kim