Using lambda conditional and pandas str.contains to lump strings

Question

Trying to learn some stuff, I'm messing around with the global shark attack database on Kaggle and I'm trying to find the best way to lump strings using a lambda function and str.contains.

Basically anywhere a string contains a phrase with skin diving e.g. 'skin diving for abalone' , in the data['Activity'] column I want to replace the activity with skin diving. (there are 92 variations for skin diving hence trying to use the lambda function)

I can return a boolean series using

data['Activity].str.contains('skin diving')

But I'm unsure how to change the value if this condition is true

My lambda function = data.apply(lambda x: 'free diving' if x.str.contains('free diving)) but i'm getting a syntax error and i'm not familiar enough with lambda functions and pandas to get it right, any help would be appreciated.

Meal

The if expression must have an else part: x if condition else y. Your lambda does not have the else part. – Hipparch

Answer 1

Instead of using a Series.str method, you can use the in operator in your lambda to test for the substring

data['activity'] = data['activity'].apply(lambda x: 'skin diving' if 'skin diving' in x else x)

Incoordination

Thanks mate, very useful. I hadn't thought about the in operator – Meal

How would I use regex against a string? I want to loop through series cells and if cell contains parenthesis ( or ) do something, else do something else – Wame

It also seems to be the case that a conditional in a lambda statement requires an else clause. – Donne

Answer 2

You could use str.contains method with np.where

In [141]: df
Out[141]:
         activity
0  free diving ok
1              ok

In [142]: df.activity = np.where(df.activity.str.contains('free diving'),
                                 'free diving', df.activity)

In [143]: df
Out[143]:
      activity
0  free diving
1           ok

Maynard