Question

假设我有一些 JSON 如下

response = {
        "totalrecords": 2,
        "data": [
            {
                "stateCd": "U.K",
                "stateName": "uttarakhand",
                "details": {
                    "id": [
                        "2312-k",
                        "2312-k"
                    ],
                    "date": [
                        "10-OCT-2019",
                        "11-OCT-2019"
                    ],
                    "icp": [
                        2233,
                        6443
                    ],
                    "icpr": [
                        3.434,
                        23.232
                    ]
                }
            },
            {
                "stateCd": "U.P",
                "stateName": "uttar pradesh",
                "details": {
                    "id": [
                        "2712-k",
                        "5412-k"
                    ],
                    "date": [
                        "10-OCT-2019",
                        "11-OCT-2019"
                    ],
                    "icp": [
                        2233,
                        6443
                    ],
                    "icpr": [
                        32.434,
                        31.232
                    ]
                }
            }
        ]
    }

我想将其转换为如下数据框

enter image description here

但是在尝试使用 pandas.json_normalize() 将其转换为数据帧时我无法达到我想要的输出

我尝试过的：

data_trunc=response['data'] # to extract data from response
pd.json_normalize(data_trunc)

enter image description here

pd.json_normalize(data_trunc,record_path=['details','id'],meta=['stateCd','stateName'])

enter image description here

但这不包括 date、icp、icpr 列

所以我尝试了不同的排列组合

    pd.json_normalize(data_trunc,record_path=[['details','id'],['date']],meta=['stateCd','stateName'])

pd.json_normalize(data_trunc,record_path=[['details','id'],['details'.'date']],meta=['stateCd','stateName'])

但遇到同样的错误 TypeError: unhashable type: 'list'

Answer 1

你需要爆炸。

pd.json_normalize(data_trunc).apply(pd.Series.explode)

<头>

	stateCd	stateName	details.id	details.date	details.icp	details.icpr
0	英国	北阿坎德邦	2312-k	10-OCT-2019	2233	3.434
0	英国	北阿坎德邦	2312-k	11-OCT-2019	6443	23.232
1	U.P	北方邦	2712-k	10-OCT-2019	2233	32.434
1	U.P	北方邦	5412-k	11-OCT-2019	6443	31.232

将嵌套的 json 转换为具有特定输出的数据帧

1 个答案: