Pandas: update content or row found with loc with the content of another columns of the same row

Question

I'm struglling concatenating some values contained in different columns of a Pandas.DataFrame.

Given a DataFrame like this:

type	maker	detailsAsDict
car	jeep	{model: '1', color: 'red'}
car	jeep	{model: '2', color: 'white'}
truck	ford	None

I want to change the value of maker where the type is car, so that the DF would look like this:

type	maker	detailsAsDict
car	jeep, model: 1	{model: '1', color: 'red'}
car	jeep, model: 2	{model: '2', color: 'white'}
truck	ford	None

To do this, I've tried using loc:

cond = df['type'] == 'car'
df.loc[cond, 'maker'] = df['maker'] + ', model:' + df['detailsAsDict']['model']

However, this results in the error KeyError: model.

So I tried the following:

def test_data(data):
    print(data['model']) 
    return data['model']

df.loc[cond, 'maker'] = test_data(df['detailsAsDict'])

But then test_data prints all None values.

Is it possible to achieve what I'm trying to do? Am I missing something?

Answer 1

You can use str accessor to access values of dicts in a column:

cond = df['type'] == 'car'
df.loc[cond, 'maker'] = df['maker'] + ', model:' + df['detailsAsDict'].str['model']

Output:

    type          maker                     detailsAsDict
0    car  jeep, model:1    {'model': '1', 'color': 'red'}
1    car  jeep, model:2  {'model': '2', 'color': 'white'}
2  truck           ford                              None

As for using test_data to assign values, print returns None, so you're really assigning None to df.loc[cond, 'maker'].