2

I have a df:

  dog1  dog2  cat1  cat2  ant1  ant2
0    1     2     3     4     5     6
1    1     2     3     4     0     0
2    3     3     3     3     3     3
3    4     3     2     1     1     0

I want to add a new column based on the following conditions:

 if   max(dog1, dog2) > max(cat1, cat2) > max(ant1, ant2) ----->   2
 elif max(dog1, dog2) > max(cat1, cat2)                   ----->   1
 elif max(dog1, dog2) < max(cat1, cat2) < max(ant1, ant2) ----->  -2
 elif max(dog1, dog2) < max(cat1, cat2)                   ----->  -1
 else                                                     ----->   0

So it should become this:

  dog1  dog2  cat1  cat2  ant1  ant2   new
0    1     2     3     4     5     6    -2
1    1     2     3     4     0     0    -1
2    3     3     3     3     3     3     0
3    4     3     2     1     1     0     2

I know how to do it with straightforward condition, but not this kind with max. What's the best way to do it?

saga
  • 736
  • 2
  • 8
  • 20

3 Answers3

1

You can use .max(axis=1) function in pandas for it:

conditions = [
       (df[['dog1','dog2']].max(axis=1) > df[['cat1','cat2']].max(axis=1)) & (df[['cat1','cat2']].max(axis=1) > df[['ant1','ant2']].max(axis=1)), 
       (df[['dog1','dog2']].max(axis=1) > df[['cat1','cat2']].max(axis=1)),
       (df[['dog1','dog2']].max(axis=1) < df[['cat1','cat2']].max(axis=1)) & (df[['cat1','cat2']].max(axis=1) < df[['ant1','ant2']].max(axis=1)), 
       (df[['dog1','dog2']].max(axis=1) < df[['cat1','cat2']].max(axis=1))]
choices = [2,1,-2,-1]
df['new'] = np.select(conditions, choices, default=0)

output:

   dog1  dog2  cat1  cat2  ant1  ant2  new
0     1     2     3     4     5     6   -2
1     1     2     3     4     0     0   -1
2     3     3     3     3     3     3    0
3     4     3     2     1     1     0    2
Ehsan
  • 12,072
  • 2
  • 20
  • 33
1

You can use apply Documentation 

def newrow(dog1,dog2,cat1,cat2,ant1,ant2):
    if max(dog1, dog2) > max(cat1, cat2) > max(ant1, ant2):
        return 2
    elif max(dog1, dog2) > max(cat1, cat2):
        return 1
    elif max(dog1, dog2) < max(cat1, cat2) < max(ant1, ant2):
        return -2
    elif max(dog1, dog2) < max(cat1, cat2):
        return -1
    return 0

df['new'] = df.apply(lambda x: newrow(*x), axis=1)

The new df will be 

  dog1  dog2  cat1  cat2  ant1  ant2  new
0     1     2     3     4     5     6   -2
1     1     2     3     4     0     0   -1
2     3     3     3     3     3     3    0
3     4     3     2     1     1     0    2
Mohnish
  • 1,010
  • 1
  • 12
  • 20
0

It seems you looking for np.maximum(). Try to find it out at numpy maximum Hope it help.

ngo huy
  • 26
  • 6