CODEWITHSUNDEEP
pythonpandaslambdagroup-by
Printable Bao
Printable Bao

Reputation: 115

Pandas groupwise row value adjustments

I wrote a lambda function to adjust a dataframe like this:

df = pd.DataFrame({'types':['A','A','A','A','A','B','B','B','B','B'],'disruptions':['N','N','Y','N','N','N','N','N','N','N'],'actions':[0,0,1,0,0,0,0,1,0,0],'status':[0,0,0,1,1,0,0,0,1,1],'status_2':[0,0,0,0,1,0,0,0,1,1]})

My goal is to convert the 'status' column into 'status_2'. The logic is: within the same 'types' group, if 'action' is labeled as 1 & 'disruptions' is 'Y', then the next row of 'status_2' will be 0. Otherwise equal to 'status'.

Here is what I have:

df['status_2'] = df.groupby('types').apply(lambda x: 0 if x['disruptions'].shift(1) == 'Y' & x['actions'].shift(1) == 1 else x['status'])

The error is: TypeError: cannot compare a dtyped[float64] array with a scalar of type[bool].

Can someone point out what I'm doing wrong here?

Upvotes: 1

Views: 88

Answers (1)

jezrael
jezrael

Reputation: 862751

I think you need create mask by groupby, convert to numpy array by values and then use Series.mask, which for True values get 0, another values are untouched:

f = lambda x: (x['disruptions'].shift(1) == 'Y') & (x['actions'].shift(1) == 1)
m = df.groupby('types').apply(f).values
print (m)
[False False False  True False False False False False False]

#changed column name for compare
df['status_2_new'] = df['status'].mask(m, 0)
print (df)
   actions disruptions  status  status_2 types  status_2_new
0        0           N       0         0     A             0
1        0           N       0         0     A             0
2        1           Y       0         0     A             0
3        0           N       1         0     A             0
4        0           N       1         1     A             1
5        0           N       0         0     B             0
6        0           N       0         0     B             0
7        1           N       0         0     B             0
8        0           N       1         1     B             1
9        0           N       1         1     B             1

Upvotes: 1

Related Questions