Change multiple boolean columns' values based on a different columns values, using a list

user_13343402 0
Viewed 43

There is a list of values

weather = ['cloudy', 'sunny']

I've got a dataframe with an old column "weather". We switched to 2 newer columns with boolean values, so all the old columns need to be accounted for.

Here is my dataframe now:

[In]
data = [['cloudy', False, False], ['sunny', False, False]]
df = pd.DataFrame(data, columns=['old', 'cloudbool', 'sunbool'])
df
[Out]
     old  cloudbool sunbool
0   cloudy  False   False
1   sunny   False   False

Desired output:

[In]
data = [['cloudy', True, False], ['sunny', False, True]]
df = pd.DataFrame(data, columns=['old', 'cloudbool', 'sunbool'])
[Out]
    old   cloudbool sunbool
0   cloudy  True    False
1   sunny   False   True

I know I could do something like what I've got below, but I've got a list of "weather types" much longer than 2.

df.loc[df['old'] == 'cloudy', ['cloudbool']] = True

I hope I conveyed that properly. Thank you

pandas boolean list python
2 Answers

Let's try str.get_dummies to create dummy indicator variables, then join it with original dataframe:

df[['old']].join(df['old'].str.get_dummies().astype(bool).add_suffix('bool'))

      old  cloudybool  sunnybool
0  cloudy        True      False
1   sunny       False       True
user_128331660
  1. I know that the get_dummies method is built for this, but another way to do this would be to create a series from list comprehension that compares the weather values in your old column to your bool column names (assuming they already exist as in your example). Then, covert it to a list in preparation for adding it to a datframe.
  2. You don't have a direct match on the names, so I have omitted the last two characters, e.g. cloudy would be clou and sunny would be sun. I don't think any weather could have a suffix > 2 characters? Again this is why this wouldn't be as robust as get_dummies. You could also make your column names match your values, e.g. cloudybool and sunnybool:
s = df.apply(lambda x: [x['old'][:-2] in col for col in df.columns[1:]], axis=1).to_list()
df1 = pd.concat([df['old'],pd.DataFrame(s, columns=df.columns[1:])], axis=1)
df1
Out[1]: 
      old  cloudbool  sunbool
0  cloudy       True    False
1   sunny      False     True
user_63667700
Related