从嵌套字典中的值构造数据框

Question

I have a list of lists of dictionaries that I'm trying to convert to a pandas DataFrame but I'm unable to use pandas.DataFrame.from_dict() because I want the value of the 'name' key to be the column header and the value of the 'duration' key to be the row value. Any suggestions on how I can make this work?

[[{'duration': 21.82, 'name': 'ABC'},
{'duration': 3.9, 'name': 'DEF'},
{'duration': 105.78, 'name': 'GHI'},
{'duration': 63.14, 'name': 'JKL'}],
[{'duration': 18.9, 'name': 'ABC'},
{'duration': 56.01, 'name': 'DEF'},
{'duration': 38.36, 'name': 'GHI'},
{'duration': 34.16, 'name': 'JKL'}]]

Desired Output:

    ABC    DEF    GHI     JKL
0  21.82   3.9   105.78   63.14
1  18.9   56.01  38.36    34.16

Answer 1

You can flatten your list of lists via itertools.chain , then pivot your dataframe. The trick here is to use an index which counts cumulatively by your grouper.

from itertools import chain

df = pd.DataFrame(list(chain.from_iterable(L)))

res = df.pivot(index=df.groupby('name').cumcount(), columns='name')
res.columns = res.columns.droplevel(0)  # remove unwanted column level

print(res)

name    ABC    DEF     GHI    JKL
0     21.82   3.90  105.78  63.14
1     18.90  56.01   38.36  34.16