如何将数组字典数组转换为 pandas dataframe？

Question

I have an array of nested dictionary:

data = {"A":"a","B":"b","ID":[{"ii":"ABC","jj":"BCD"},{"ii":"AAC","jj":"FFD"}],"Finish":"yes"}

I used,

res = pd.DataFrame.from_dict(data , orient='index')

But the ID is still returned as list of dictionary.

A  B      ID                                              Finish
a  b  [{"ii":"ABC","jj":"BCD"},{"aa":"AAC","bb":"FFD"}]    yes

But I want everything to be converted to df. Not sure how to do it. Kindly help.

Expected Output:

A  B  ID.ii  ID.jj   Finish
a  b   ABC    BCD      yes
a  b   AAC    FFD      yes

Answer 1

You can achieve this using pandas json_normalize

df = pd.json_normalize(data, meta=['A', 'B'], record_path=['ID'], record_prefix="ID.")

Output

  ID.ii ID.jj  A  B
0   ABC   BCD  a  b
1   AAC   FFD  a  b

record_path - will be used to flatten the specific key record_prefix - is added as a column prefix meta - is the columns that needs to be preserved without flattening

Refer the documentation for examples

Answer 2

To achieve this without using json_normalize , you can pre-process the input like this-

data = {"A":"a","B":"b","ID":[{"ii":"ABC","jj":"BCD"},{"ii":"AAC","jj":"FFD"}],"Finish":"yes"}
op = {}

for i in data:
    if isinstance(data[i], list):
        for j in data[i]:
            for k in j:
                tmp = str(i)+"."+str(k)
                if tmp not in op:
                    op[tmp] = [j[k]]
                else:
                    op[tmp].append(j[k])
    else:
        op[i] = data[i]

        
>>> data
{'A': 'a', 'B': 'b', 'ID': [{'ii': 'ABC', 'jj': 'BCD'}, {'ii': 'AAC', 'jj': 'FFD'}], 'Finish': 'yes'}
>>> op
{'A': 'a', 'B': 'b', 'ID.ii': ['ABC', 'AAC'], 'ID.jj': ['BCD', 'FFD'], 'Finish': 'yes'}

After this you can directly use

>>> pd.DataFrame(op)

   A  B ID.ii ID.jj Finish
0  a  b   ABC   BCD    yes
1  a  b   AAC   FFD    yes