Python : How to use if elif for rows in a dataframe

Question

I have a dataframe and i want to add a columns using if elif condition on the rows of the table. I am using if elif statement but that is not working. Can we not use the conditional statememt for a data frame?

Here is my code:

import pandas as pd
df = pd.DataFrame({'c1': ['a', 'a', 'q', 'a'],
              'c2': ['b', 'e', 'b', 'f'],
               'c3': ['c', 'f', 'c', 'd']})

if [(df['c1']=='a') & (df['c2']=='b')]:
    df['q']= df['c1'] + '+' + df['c2']
elif (df['c1']=='a' & df['c2']=='e'):
    df['q'] = df['c1'] + '*' + df['c2']
else:
    df['q'] = df['c1'] + '-' + df['c2']

The new column 'q' has contents: 'a+b', 'a+e', 'q+b', 'a+f'

While i want it as: 'a+b', 'a*e', 'q-b', 'a-f'

@AnkurSharma see https://stackoverflow.com/a/21415990/2689986 — shad0w_wa1k3r, Aug 14 '17 at 12:05
'and' doesn't work. It is showing an error. "The truth value of a Series is ambiguous. Use a.empty, a.bool(), a.item(), a.any() or a.all()." — sky_bird, Aug 14 '17 at 12:06
You might also want to consider creating a function to assign the value of `df['q']` then apply that function to each row. That will make the implementation easier to read rather than having this long if-else statement of column references. — vielkind, Aug 14 '17 at 12:08

score 2 · Accepted Answer · answered Aug 14 '17 at 12:18

2

Use numpy.select what is better readability form of multiple nested np.where:

m1 = (df['c1']=='a') & (df['c2']=='b')
m2 = (df['c1']=='a') & (df['c2']=='e')

a1 = df['c1'] + '+' + df['c2']
a2 = df['c1'] + '*' + df['c2']
a3 = df['c1'] + '-' + df['c2']

df['q'] = np.select([m1, m2], [a1, a2], default=a3)
print (df)

  c1 c2 c3    q
0  a  b  c  a+b
1  a  e  f  a*e
2  q  b  c  q-b
3  a  f  d  a-f

answered Aug 14 '17 at 12:18

jezrael

822,522
95
1,334
1,252

1

It helped me..thanx!! – sky_bird Aug 15 '17 at 09:03
Glad can help, nice day! – jezrael Aug 15 '17 at 09:13

score 1 · Answer 2 · answered Aug 14 '17 at 12:17

Vectorized version of if statements is np.where. I assigned the conditions and the possible outcomes to variables to improve readability because with nested np.wheres it may become hard to follow.

cond1 = (df['c1']=='a') & (df['c2']=='b')
cond2 = (df['c1']=='a') & (df['c2']=='e')

case1 = df['c1'] + '+' + df['c2']
case2 = df['c1'] + '*' + df['c2']
case3 = df['c1'] + '-' + df['c2']

df['q'] = np.where(cond1, case1, np.where(cond2, case2, case3))

df
Out: 
  c1 c2 c3    q
0  a  b  c  a+b
1  a  e  f  a*e
2  q  b  c  q-b
3  a  f  d  a-f

Python : How to use if elif for rows in a dataframe

2 Answers2