Python：字典到 pandas dataframe

Question

I would like to change the dictionary to pandas dataframe.

data = {u'Diluted Normalized EPS': [{u'date': u'2020-01-03', u'value': u'-0.446810'}, {u'date': u'2019-10-04', u'value': u'-0.765540'}, {u'date': u'2019-06-28', u'value': u
'-0.574240'}, {u'date': u'2019-03-29', u'value': u'-2.063700'}, {u'date': u'2018-12-28', u'value': u'-0.841380'}], u'Net Income Before Extra. Items': [{u'date': u'2020-01-03', u'value': u'-139.000000'}, {u'date': u'2019-10-04', u'value': u'-276.000000'}, {u'date': u'2019-06-28', u'value': u'-185.000000'}, {u'date': u'2019-03-29', u'value': u'-652.000000'}, {u'date': u'2018-12-28', u'value': u'-257.000000'}]}

I would like to convert to Pandas dataframe as below

         date  Diluted Normalized EPS  Net Income Before Extra. Items
0  2020-01-03  -0.446810                -139.000000
1  2019-10-04  -0.765540                -276.000000
2  2019-06-28  -0.574240                -185.000000
3  2019-03-29  -2.063700                -652.000000
4  2018-12-28  -0.841380                -257.000000

pd.DataFrame(data) unable to return the desired results

Answer 1

Use nested dictionary comprehension with DataFrame constructor:

d = {k:{x['date']: x['value'] for x in v} for k, v in data.items()}
df = pd.DataFrame(d).rename_axis('date').reset_index()
print(df)
         date Diluted Normalized EPS Net Income Before Extra. Items
0  2020-01-03              -0.446810                    -139.000000
1  2019-10-04              -0.765540                    -276.000000
2  2019-06-28              -0.574240                    -185.000000
3  2019-03-29              -2.063700                    -652.000000
4  2018-12-28              -0.841380                    -257.000000

Answer 2

Run the below code:

df = pd.DataFrame.from_dict(data).apply(
    lambda col: pd.DataFrame(col.tolist()).set_index('date').value)\
    .reset_index()

Steps:

1. pd.DataFrame.from_dict(data) - creates a temporary DataFrame. It has 2 columns (keys of your dictionary). Each cell contains a dictionary, containing 2 keys ( date and value ).

2. pd.DataFrame(col.tolist()) - the first part of the lambda function - converts the current column into another temporary DataFrame, with 2 columns ( date and value ).

3. set_index('date') - changes date column into the index.

4. value - takes only value column from the above DataFrame.

5. apply(...) - application of the above lambda function converts the first temporary DataFrame into almost exactly what we want, ie 2 required columns with date as the index.

6. reset_index() - converts the index into a regular column.