Python 在大熊猫中分解和解包列

假设我有一个如下结构的数据，如何分解包含列表的列，然后解压分解的列

来源：

d = { 
    "_id" : "5f2", 
    "connId" : 128, 
    "hospitalList" : [
        {
            "hospitalId" : 29, 
            "boardId" : 1019, 
            "siteId" : 1
        }, 
        {
            "hospitalId" : 3091, 
            "boardId" : 2163, 
            "siteId" : 382
        },
      {
            "hospitalId" : 28, 
            "boardId" : 1017, 
            "siteId" : 5
        }]
    }

root = pd.json_normalize(d)
nested_cols = [i for i in root.columns if isinstance(root[i][0], list)]
l = [root.drop(nested_cols,1),]
for i in nested_cols:
    l.append(pd.json_normalize(d, record_path=i))

output = pd.concat(l, axis=1)
print(output)

                        _id      connId  hospitalId       boardId  siteId
0                       5f2       128.0          29         1019       1
1                       NaN         NaN        3091         2163     382
2                       NaN         NaN          28         1017       5

                        _id      connId  hospitalId      boardId  siteId
0                       5f2       128.0          29         1019       1
1                       5f2       128.0        3091         2163     382
2                       5f2       128.0          28         1017       5

代码：

d = { 
    "_id" : "5f2", 
    "connId" : 128, 
    "hospitalList" : [
        {
            "hospitalId" : 29, 
            "boardId" : 1019, 
            "siteId" : 1
        }, 
        {
            "hospitalId" : 3091, 
            "boardId" : 2163, 
            "siteId" : 382
        },
      {
            "hospitalId" : 28, 
            "boardId" : 1017, 
            "siteId" : 5
        }]
    }

root = pd.json_normalize(d)
nested_cols = [i for i in root.columns if isinstance(root[i][0], list)]
l = [root.drop(nested_cols,1),]
for i in nested_cols:
    l.append(pd.json_normalize(d, record_path=i))

output = pd.concat(l, axis=1)
print(output)

                        _id      connId  hospitalId       boardId  siteId
0                       5f2       128.0          29         1019       1
1                       NaN         NaN        3091         2163     382
2                       NaN         NaN          28         1017       5

                        _id      connId  hospitalId      boardId  siteId
0                       5f2       128.0          29         1019       1
1                       5f2       128.0        3091         2163     382
2                       5f2       128.0          28         1017       5

实际结果：

d = { 
    "_id" : "5f2", 
    "connId" : 128, 
    "hospitalList" : [
        {
            "hospitalId" : 29, 
            "boardId" : 1019, 
            "siteId" : 1
        }, 
        {
            "hospitalId" : 3091, 
            "boardId" : 2163, 
            "siteId" : 382
        },
      {
            "hospitalId" : 28, 
            "boardId" : 1017, 
            "siteId" : 5
        }]
    }

root = pd.json_normalize(d)
nested_cols = [i for i in root.columns if isinstance(root[i][0], list)]
l = [root.drop(nested_cols,1),]
for i in nested_cols:
    l.append(pd.json_normalize(d, record_path=i))

output = pd.concat(l, axis=1)
print(output)

                        _id      connId  hospitalId       boardId  siteId
0                       5f2       128.0          29         1019       1
1                       NaN         NaN        3091         2163     382
2                       NaN         NaN          28         1017       5

                        _id      connId  hospitalId      boardId  siteId
0                       5f2       128.0          29         1019       1
1                       5f2       128.0        3091         2163     382
2                       5f2       128.0          28         1017       5

预期结果：

d = { 
    "_id" : "5f2", 
    "connId" : 128, 
    "hospitalList" : [
        {
            "hospitalId" : 29, 
            "boardId" : 1019, 
            "siteId" : 1
        }, 
        {
            "hospitalId" : 3091, 
            "boardId" : 2163, 
            "siteId" : 382
        },
      {
            "hospitalId" : 28, 
            "boardId" : 1017, 
            "siteId" : 5
        }]
    }

root = pd.json_normalize(d)
nested_cols = [i for i in root.columns if isinstance(root[i][0], list)]
l = [root.drop(nested_cols,1),]
for i in nested_cols:
    l.append(pd.json_normalize(d, record_path=i))

output = pd.concat(l, axis=1)
print(output)

                        _id      connId  hospitalId       boardId  siteId
0                       5f2       128.0          29         1019       1
1                       NaN         NaN        3091         2163     382
2                       NaN         NaN          28         1017       5

                        _id      connId  hospitalId      boardId  siteId
0                       5f2       128.0          29         1019       1
1                       5f2       128.0        3091         2163     382
2                       5f2       128.0          28         1017       5

---

这会输出您想要的内容

root = pd.json_normalize(d)
nested_cols = [i for i in root.columns if isinstance(root[i][0], list)]
l = [root.drop(nested_cols,1),]
for i in nested_cols:
    l.append(pd.json_normalize(d, record_path=i))

output = pd.concat(l, axis=1)

output.fillna(method='ffill', inplace=True)

---

但是，不幸的是，我不知道在什么情况下您将使用代码，和/或您是否必须进行调整。

试试这个：

output.fillna（method='ffill'，inplace=True）

是否有一种方法可以使其动态化，而无需像我这样明确指定列名？我认为r-初学者的评论解决了您的问题，如果他不打算自己做一个，我会将其插入我的答案中。：）