Creating a Pandas Dataframe from a list of dictionaries, parsing

Question

I have a Python list of dictionaries, with string and integer values.

I would like to import this into a Pandas DataFrame. My first thought was to manipulate the list of dictionaries into one big dictionary, and then import this into a Pandas DataFrame.

The "one big dictionary" would be

dict_countries = { 'countries':       [],
                   'pop':             [],
                   'capital_city':    [],
                   'national_anthem': [] }

And then I could use

for dictionary in list_countries:
    dict_countries['countries'].append(dictionary['country'])
    dict_countries['pop'].append(dictionary['population'])
    dict_countries['capital_city'].append(dictionary['capital'])
    dict_countries['national_anthem'].append(dictionary['anthem'])

However, I am worried that this is a bad idea. A dictionary of lists is a bit fragile: if any of the lists get out of synch the whole thing becomes a mess.

How could I skip the middle step and immediately parse my list of dictionaries into a pandas DataFrame?

Answer 1

A DataFrame can be constructed directly from a list of dictionaries:

In [100]: pd.DataFrame([{'foo':1, 'bar':2}, {'foo':3,'baz':4}])
Out[100]: 
   bar  baz  foo
0    2  NaN    1
1  NaN    4    3