Reputation: 1967
I am trying to create a conditional column in pandas. Here is what the dataframe looks like.
data = [{"owner" : "john", "dog" : 'magie', "dog_is_fluffy" : 1},
{"owner" : "john", "dog" : 'stellar', "dog_is_fluffy" : 0},
{"owner" : "lisa", "dog" : 'mollie' , "dog_is_fluffy" : 0},
{"owner" : "lisa", "dog" : 'rex', "dog_is_fluffy" : 0},
{"owner" : "john", "dog" : 'luns', "dog_is_fluffy" : 1}]
df = pd.DataFrame(data)
As you can see, my data shows dogs and their owners. We also know if the dog is fluffy. I want to create two columns fluffy_dogs_owned
and owner_has_fluffy_dog
.
The result I am looking for is:
data_result = [{"owner" : "john", "dog" : 'magie', "dog_is_fluffy" : 1, "fluffy_dogs_owned" : 2, "owner_has_fluffy_dog" : 1},
{"owner" : "john", "dog" : 'stellar', "dog_is_fluffy" : 0, "fluffy_dogs_owned" : 2, "owner_has_fluffy_dog" : 1},
{"owner" : "lisa", "dog" : 'mollie' , "dog_is_fluffy" : 0, "fluffy_dogs_owned" : 0, "owner_has_fluffy_dog" : 0},
{"owner" : "lisa", "dog" : 'rex', "dog_is_fluffy" : 0, "fluffy_dogs_owned" : 0, "owner_has_fluffy_dog" : 0},
{"owner" : "john", "dog" : 'luns', "dog_is_fluffy" : 1, "fluffy_dogs_owned" : 2, "owner_has_fluffy_dog" : 1}]
df_result = pd.DataFrame(data_result)
I thought about using df.groupby()
and np.where
but I can't make it work so far. Any ideas?
Upvotes: 1
Views: 78
Reputation: 862791
Use GroupBy.transform
for return Series
with same size like original Dataframe with sum
and then compare column for not equal by Series.ne
with casting to integer
df['fluffy_dogs_owned'] = df.groupby('owner')['dog_is_fluffy'].transform('sum')
df['owner_has_fluffy_dog'] = df['fluffy_dogs_owned'].ne(0).astype(int)
Or with Series.clip
:
df['owner_has_fluffy_dog'] = df['fluffy_dogs_owned'].clip(upper=1)
print (df)
dog dog_is_fluffy owner fluffy_dogs_owned owner_has_fluffy_dog
0 magie 1 john 2 1
1 stellar 0 john 2 1
2 mollie 0 lisa 0 0
3 rex 0 lisa 0 0
4 luns 1 john 2 1
Upvotes: 2