Pandas/Numpy Multiple if statement with and/or operators

Question

I have quite a complex If statement that I would like to add as a column in my pandas dataframe. In the past I've always used numpy.select for this type of problem, however I wouldn't know how to achieve that with a multi-line if statement.

I was able to get this in Excel:

=IF(sum1=3,IF(AND(col1=col2,col2=col3),0,1),IF(sum1=2,IF(OR(col1=col2,col2=col3,col1=col3),0,1),IF(sum1=1,0,1)))

and write it in Python just as a regular multi-line 'if statement', just want to find out if there is a far cleaner way of presenting this.

if df['sum1'] == 3:
  if df['col1'] == df['col2'] and df['col2'] == df['col3']:
    df['verify_col'] = 0
  else:
    df['verify_col'] = 1
elif df['sum1'] == 2:
  if df['col1'] == df['col2'] or df['col2'] == df['col3'] or df['col1'] == df['col3']:
    df['verify_col'] = 0
  else:
    df['verify_col'] = 1
elif df['sum1'] == 1:
  df['verify_col'] = 0
else:
  df['verify_col'] = 1

Here's some sample data:

df = pd.DataFrame({
    'col1': ['BMW', 'Mercedes Benz', 'Lamborghini', 'Ferrari', null],
    'col2': ['BMW', 'Mercedes Benz', null, null, 'Tesla'],
    'col3': ['BMW', 'Mercedes', 'Lamborghini', null, 'Tesla_'],
    'sum1': [3, 3, 2, 1, 2]
})

I want a column which has the following results:

'verify_col': [0, 1, 0, 0, 1]

It basically checks whether the columns match for those that have values in them and assigns a 1 or a 0 for each row. 1 meaning they are different, 0 meaning zero difference.

I would be tempted to use .apply for this. Are you familiar with that? It calls a function that you provide on every row, and the result can be assigned to a new column. — Tim Roberts
– Tim Roberts, Commented Oct 25, 2022 at 5:26

jezrael · Accepted Answer · 2022-10-25 05:37:26Z

4

Use numpy.where with chain mask with | for bitwise OR - if no match any conditions is created 1:

m1 = (df['sum1'] == 3)
m2 = (df['col1'] == df['col2']) & (df['col2'] == df['col3'])
m3 = (df['sum1'] == 2)
m4 = (df['col1'] == df['col2']) | (df['col2'] == df['col3']) | (df['col1'] == df['col3'])
m5 = df['sum1'] == 1

df['verify_col'] = np.where((m1 & m2) | (m3 & m4) | m5, 0, 1)

If need None if no match any conditions:

df['verify_col'] = np.select([(m1 & m2) | (m3 & m4) | m5,
                              (m1 & ~m2) | (m3 & ~m4) | ~m5], 
                             [0,1], default=None)


print (df)
            col1           col2         col3  sum1  verify_col
0            BMW            BMW          BMW     3           0
1  Mercedes Benz  Mercedes Benz     Mercedes     3           1
2    Lamborghini            NaN  Lamborghini     2           0
3        Ferrari            NaN          NaN     1           0
4            NaN          Tesla       Tesla_     2           1

edited Oct 25, 2022 at 5:37

answered Oct 25, 2022 at 5:26

jezrael

868k103 gold badges1.4k silver badges1.3k bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

jezrael Over a year ago

@DeepakTripathi - I have typo, now correct

sammywemmy · Accepted Answer · 2022-10-25 06:07:32Z

One option is with case_when from pyjanitor:

# pip install pyjanitor
import pandas as pd
import janitor

(df
.case_when(
# condition, result
df.sum1.eq(3) & df.col1.eq(df.col2) & df.col2.eq(df.col3), 0,
df.sum1.eq(3), 1,
df.sum1.eq(2) & (df.col1.eq(df.col2) | df.col2.eq(df.col3) | df.col1.eq(df.col3)), 0,
df.sum1.eq(2), 1,
df.sum1.eq(1), 0,
1, # default
column_name='verify_col')
)

            col1           col2         col3  sum1  verify_col
0            BMW            BMW          BMW     3           0
1  Mercedes Benz  Mercedes Benz     Mercedes     3           1
2    Lamborghini           None  Lamborghini     2           0
3        Ferrari           None         None     1           0
4           None          Tesla       Tesla_     2           1

Of course, you can do this with np.select:

conditions = [df.sum1.eq(3) & df.col1.eq(df.col2) & df.col2.eq(df.col3), 
              df.sum1.eq(3), 
              df.sum1.eq(2) & (df.col1.eq(df.col2) | df.col2.eq(df.col3) | 
              df.col1.eq(df.col3)), 
              df.sum1.eq(2), 
              df.sum1.eq(1)]

results = [0,1,0,1,0]

outcome = np.select(conditions, results, default=1)
df.assign(verify_col = outcome)

            col1           col2         col3  sum1  verify_col
0            BMW            BMW          BMW     3           0
1  Mercedes Benz  Mercedes Benz     Mercedes     3           1
2    Lamborghini           None  Lamborghini     2           0
3        Ferrari           None         None     1           0
4           None          Tesla       Tesla_     2           1

Raibek · Accepted Answer · 2022-10-25 05:38:42Z

0

df['verify_col'] = (~(((df["col1"] == df["col2"]) | df["col1"].isna() | df["col2"].isna()) & ((df["col2"] == df["col3"]) | df["col2"].isna() | df["col3"].isna()))).astype(int)

answered Oct 25, 2022 at 5:38

Raibek

5783 silver badges6 bronze badges

Collectives™ on Stack Overflow

Pandas/Numpy Multiple if statement with and/or operators

3 Answers 3

1 Comment

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

1 Comment

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related