How to explode a string column with dict-like format in pandas-CodePudding

I have a columns in a dataframe like this:

                               column_name
{"trials": {"value": ["8"]}, "results": {"value": "malfuction details."}, "trials_nic": {"value": ["7"]}, "custom_cii": {"value": ["yes"]}}

When I apply type to this column I get "str"

df['column_name'].apply(type)

output: <class 'str'>

How can I flat this column to get each key:value pair in a new column?

CodePudding user response：

The canonical way to explode a column of dictionaries is:

new_df = pd.DataFrame(df['column_name'].tolist())

However, you probably want the additional step of extracting your values from the dictionaries they are contained within:

new_df = new_df.applymap(lambda d: d['value'])

Combining both steps into a single one-liner:

new_df = pd.DataFrame(df['column_name'].tolist()).applymap(lambda d: d['value'])

Which gives you what I assume is your desired output:

CodePudding user response：

you can use:

dictt = {"trials": {"value": ["8"]}, "results": {"value": "malfuction details."}, "trials_nic": {"value": ["7"]}, "custom_cii": {"value": ["yes"]}}

df=pd.DataFrame(data={'column_name':[dictt]})
print(df)
'''

    column_name
0   {'trials': {'value': ['8']}, 'results': {'value': 'malfuction details.'}, 'trials_nic': {'value': ['7']}, 'custom_cii': {'value': ['yes']}}

'''

df=df['column_name'].apply(pd.Series).T
df=df[0].apply(pd.Series).explode('value').T
print(df)
'''
        trials  results             trials_nic  custom_cii
value   8       malfuction details. 7           yes

'''