the count method in python can not ignore the empty string

Question

import pandas as pd
df = pd.DataFrame({'x':[1,2,1,2,1,3,2],'y':[34,23,23,65,45,12,28],'z':['a','b','a','','a','c','b']})
df.groupby('x').z.count().reset_index()

  x  z
0  1  3
1  2  3
2  3  1

But this is what I don't want, the empty string should not be in the count, so what I want is

  x  z
0  1  3
1  2  2
2  3  1

what should I do?

Answer 1

in python the empty string is not considered null, you can replace it as null and do the same.

df['z']=df['z'].replace({'':np.NAN})
df.groupby('x').z.count().reset_index()

Answer 2

Using replace:

df.replace({'z':''},np.nan).groupby('x').z.count().reset_index()
    x   z
0   1   3
1   2   2
2   3   1

Or

df.replace({'z':''},np.nan).groupby('x',as_index=False).z.count()