2

I have a pandas df that contains a column of positive, negative nos and zeros. I wanted to crate another column which is 1 if no is > 0, -1 if no is < 0 and 0 if the number is 0.

I am trying to do this using a for loop for each row but it is taking too long. I wanted to know if there was a faster way to do this. I also wanted to know if the same logic could be extended to positive and negative timedelta objects.
Thank you.

My final df should look like this:

df = pd.DataFrame({'a':[1, 2, -1, 0, -2], 'b':[1, 1, -1, 0, -1]})

     a   b
0    1   1
1    2   1
2   -1  -1
3    0   0
4   -2  -1

where b is the col to assign based on values of a

Yuca
  • 6,010
  • 3
  • 22
  • 42
Moshee
  • 544
  • 1
  • 3
  • 16

2 Answers2

3

Here is one way numpy sign 

np.sign(df.a)
Out[118]: 
0    1
1    1
2   -1
3    0
4   -1
Name: a, dtype: int64
df['b'] = np.sign(df.a)
BENY
  • 317,841
  • 20
  • 164
  • 234
  • Thanks. Is there any way to do this for timedelta objects. I tried it but it doesn't work – Moshee Aug 02 '19 at 15:33
  • 1
    @Moshee `np.sign( (df.a / np.timedelta64(1, 'D')).astype(int))` – BENY Aug 02 '19 at 15:35
  • @Moshee more info https://stackoverflow.com/questions/25646200/python-convert-timedelta-to-int-in-a-dataframe/42247228 – BENY Aug 02 '19 at 15:37
2

try using np.where and provide conditions

import numpy as np

df['b']= np.where(df['a']>0,1,
         np.where(df['a']<0,-1,0))

     a   b
0    1   1
1    2   1
2   -1  -1
3    0   0
4   -2  -1

Solution by @rafaelc

m1= df['a'] >0
m2= df['a'] <0


df['b'] = np.select([m1, m2],
                    [ 1, -1], 
                    default=0)
Community
  • 1
  • 1
tawab_shakeel
  • 3,701
  • 10
  • 26