pandas where

1# replaces values with other where condition is False
2DataFrame.where(cond, other=nan, inplace=False, axis=None, level=None, errors='raise', try_cast=False)
3
4import pandas as pd 
5df = pd.DataFrame({'values':[1,2,3,4]})
6
7df.where(df['values'] % 2 == 0, -1) # output : [-1, 2, -3, 4]

1In [20]: df
2Out[20]:
3   A  B  C
40  1  2  1
51  2  3  0
62  3  4  0
73  4  5  1
8
9# Note that df.C is a mask
10# Note that this is np.where, not df.where. That is different.
11In [21]: df['D'] = np.where(df.C, df.A, df.B)
12
13In [22]: df
14Out[22]:
15   A  B  C  D
160  1  2  1  1
171  2  3  0  3
182  3  4  0  4
193  4  5  1  4
20