CODEWITHSUNDEEP
pythonpython-3.xpandaslistdictionary
Simon
Simon

Reputation: 71

Python dictionary of list to Pandas dataframe

I am trying to transform a dictionary of lists (looks like a dictionary of dictionary, but is unfortunately a dictionary of lists) into a dataframe. I want to have the column-names from the list objects. So far i found a way to to turn the dictionary into a data frame, but the columns don't have the appropriate name and the values still contain the column names. 

user_dict = {'Category 1': ['att_1: 1', 'att_2:  whatever'],
                  'Category 2': ['att_1 : 23', 'att_2 : another']}

res = pd.DataFrame.from_dict(user_dict, orient='index')
res.columns = [f'SYN{i+1}' for i in res]

Example Output: 

                att_1 | att_2 

Category_1       1 | whatever

Category_1       23 | another

I was thinking at using unlist or regex, but I am not sure where to input that. Any help much appreciated! Thank you

Edit: my unlist attemp ended here:

pd.DataFrame.from_dict({(i,j): to_dict(unlist(user_dict[i][j])) 
                       for i in user_dict.keys() 
                       for j in user_dict[i].keys()},
                   orient='index')

Upvotes: 1

Views: 1666

Answers (1)

jpp
jpp

Reputation: 164673

You can use a dictionary comprehension to restructure your input into a dictionary of dictionaries. Then use from_dict with orient='index':

user_dict = {'Category 1': ['att_1: 1', 'att_2:  whatever'],
             'Category 2': ['att_1 : 23', 'att_2 : another']}

d = {k: dict(map(str.strip, x.split(':')) for x in v) for k, v in user_dict.items()}

df = pd.DataFrame.from_dict(d, orient='index')

df['att_1'] = pd.to_numeric(df['att_1'])

print(df)

           att_1     att_2
Category 1     1  whatever
Category 2    23   another

As above, you will need to then convert series to numeric as appropriate.

Upvotes: 1

Related Questions