Conditionally calculated column for a Pandas DataFrame

Question

I have a calculated column in a Pandas DataFrame which needs to be assigned base upon a condition. For example:

if(data['column_a'] == 0):
    data['column_c'] = 0
else:
    data['column_c'] = data['column_b']

However, that returns an error:

ValueError: The truth value of a Series is ambiguous. Use a.empty, a.bool(), a.item(), a.any() or a.all().

I have a feeling this has something to do with the fact that is must be done in a matrix style. Changing the code to a ternary statement doesn't work either:

data['column_c'] = 0 if data['column_a'] == 0 else data['column_b']

Anyone know the proper way to achieve this? Using apply with a lambda? I could iterate via a loop, but I'd rather keep this the preferred Pandas way.

Solution

You can do:

data['column_c'] = data['column_a'].where(data['column_a'] == 0, data['column_b'])

this is vectorised your attempts failed because the comparison with if doesn't understand how to treat an array of boolean values hence the error

Example:

In [81]:
df = pd.DataFrame(np.random.randn(5,3), columns=list('abc'))
df

Out[81]:
          a         b         c
0 -1.065074 -1.294718  0.165750
1 -0.041167  0.962203  0.741852
2  0.714889  0.056171  1.197534
3  0.741988  0.836636 -0.660314
4  0.074554 -1.246847  0.183654

In [82]:
df['d'] = df['b'].where(df['b'] < 0, df['c'])
df

Out[82]:
          a         b         c         d
0 -1.065074 -1.294718  0.165750 -1.294718
1 -0.041167  0.962203  0.741852  0.741852
2  0.714889  0.056171  1.197534  1.197534
3  0.741988  0.836636 -0.660314 -0.660314
4  0.074554 -1.246847  0.183654 -1.246847