How to wrap cell value in a dict maintaining value without using apply-CodePudding

I have a Pandas dataframe with a colum with some string and some NaN values.

df['column'] = ['a', 'b', NaN, 'c']

What I want to achieve is to transform the value in dicts, maintaining the original value:

df['column'] = [{'key': 'a'}, {'key': 'b'} , {'key': None}, {'key': 'c'}]

I successfully got it with an apply:

df['column'] = df['column'].apply(lambda value: {"value": None} if type(value) == pd._libs.missing.NAType or pd.isna(value) or pd.isnull(value) or value == '' else {"value": value})

But i'd like to reach the same result using the vectorized approach if possibile.

CodePudding user response：

Vectorization is not possible since you are using complex object types like dictionaries having said that here is a slightly more efficient and concise approach than apply:

df['column'].replace(['', np.nan], None).map(lambda k: {'key': k})

Result

0     {'key': 'a'}
1     {'key': 'b'}
2    {'key': None}
3     {'key': 'c'}
Name: column, dtype: object