If-else conditional assignment in pandas

Question

I want to assign values to a column depending on the values of an already-existing column. This code works, but I would like to do it not-in-place, perhaps using assign or apply.

If this could be done in one step it would also avoid the implicit conversion from int to float that occurs below.

I've included my attempt using assign, which raises a ValueError.

import pandas as pd

original = pd.DataFrame({'col': ['a', 'b', 'c']})

d = original.copy()
d.loc[d.col.isin(['b', 'x']), 'new'] = 1
d.loc[~d.col.isin(['b', 'x']), 'new'] = 99
d

# :   col   new
# : 0   a  99.0
# : 1   b   1.0
# : 2   c  99.0

# original.assign(new=lambda x: (1 if x.col.isin(['b', 'x']) else 99)) # ValueError

Answer 1

You can use numpy.where():

import numpy as np
original["new"] = np.where(original["col"].isin(["b", "x"]), 1, 99)
print(original)
#  col  new
#0   a   99
#1   b    1
#2   c   99