CODEWITHSUNDEEP
pythonpandas
Thanos
Thanos

Reputation: 2572

Groupby with conditions in pandas

I have a pd.DataFrame that looks like this:

In [30]: df
Out[30]: 
        DATES  UID      A
0  2014-01-01    1  False
1  2014-01-02    2  False
2  2014-01-03    3   True
3  2014-01-04    4   True
4  2014-01-05    5  False
5  2014-01-06    6   True
6  2014-01-07    1  False
7  2014-01-08    2  False
8  2014-01-09    3  False
9  2014-01-10    2  False
10 2014-01-11    3  False
11 2014-01-12    4  False
12 2014-01-13    5  False
13 2014-01-14    3  False
14 2014-01-15    1  False

and I would like to find a way to:

  1. Order by DATES ASC
  2. Group by UID
  3. Filter out all UID's where the first entry (per UID) has 'A' == False

The desired output would look like this:

In [30]: df
Out[30]: 
    DATES  UID      A
0  2014-01-03    3   True
1  2014-01-04    4   True
2  2014-01-06    6   True
3  2014-01-09    3  False
4  2014-01-11    3  False
5  2014-01-12    4  False
6  2014-01-14    3  False

Any ideas very much appreciated, thanks!

Upvotes: 0

Views: 88

Answers (1)

jezrael
jezrael

Reputation: 862431

It looks like need first sort_values and then filter:

df.sort_values(by='DATES', inplace=True)
df = df.groupby('UID', sort=False).filter(lambda x: x.A.iloc[0] == True)
print (df)
        DATES  UID      A
2  2014-01-03    3   True
3  2014-01-04    4   True
5  2014-01-06    6   True
8  2014-01-09    3  False
10 2014-01-11    3  False
11 2014-01-12    4  False
13 2014-01-14    3  False

Upvotes: 1

Related Questions