CODEWITHSUNDEEP
pythonpandasgroup-by
k88074
k88074

Reputation: 2164

How to do Pandas groupby with own criteria?

Consider the following dataframe 

d = {'N': [1, 2, 1, 2], 'NV': [50, 50,200,200], 'VAL' : [100,0,10,0]}
df = pd.DataFrame(data = d)

giving

   N  NV  VAL
0  1  50  100
1  2  50    0
2  1  200  10
3  2  200   0

I would like to groupby in order to obtain, for each NV, the number of VAL entries which are strictly positive. The result I wish to obtain is the following 

   NV  VAL
0  50   1
1  200  1

Any suggestions?

Upvotes: 1

Views: 40

Answers (3)

mahmoud zridi
mahmoud zridi

Reputation: 21

try this

df['VAL'] = df.loc[df['VAL']>0].groupby(df['NV']).transform('count')
df.dropna(axis=0,inplace=True)

Upvotes: 0

luigigi
luigigi

Reputation: 4215

Try:

df[df.VAL>0].groupby('NV').size().reset_index().rename(columns={0:'VAL'})

    NV  VAL
0   50    1
1  200    1

Upvotes: 0

distracted-biologist
distracted-biologist

Reputation: 808

df.query("VAL > 0").groupby("NV").size()

or

df.groupby("NV").VAL.agg(lambda y: (y > 0).sum())

Upvotes: 3

Related Questions