Python 熊猫：将数据帧与嵌套数组组合或合并JSON输出

我使用一个标准的数据帧，并使用嵌套数组创建汇总数据的各种子集数据帧。然后，我需要以一种能够提供预期JSON输出的方式组合子集数据帧。（我用MaxU的答案格式化了大部分代码；）

我的标准数据帧的前几行（如果必要，我可以给出这个示例中的所有58行）：df

在这里，我使用以下Python：

 PAFF_df = pd.DataFrame(df.groupby(['PRI_DEP','PRI_AFF'])['ID'].nunique().unstack().reset_index().fillna(0))
 LOA_df = pd.DataFrame(df.groupby(['PRI_DEP','LOA'])['ID'].nunique().unstack().reset_index().fillna(0))
 ST_df = pd.DataFrame(df.groupby(['PRI_DEP','STATE'])['ID'].nunique().unstack().reset_index().fillna(0))

 Nested_PAFF_df = (PAFF_df.groupby(['PRI_DEP'], as_index=True)
      .apply(lambda x: x[['A','E','F','L','M','T']].to_dict('r'))
      .reset_index()
      .rename(columns={0:'Primary_Affiliation'}))

 Nested_LOA_df = (LOA_df.groupby(['PRI_DEP'], as_index=True)
      .apply(lambda x: x[['Basic','Blue','Bronze','Invalid','UFM']].to_dict('r'))
      .reset_index()
      .rename(columns={0:'LOA'}))

 Nested_ST_df = (ST_df.groupby(['PRI_DEP'], as_index=True)
      .apply(lambda x: x[['A','E']].to_dict('r'))
      .reset_index()
      .rename(columns={0:'STATE'}))

这为我提供了适当的嵌套json，使用：.to_json（orient='records'））

主要附属机构：

[{"PRI_DEP":" ","Primary_Affiliation":[{"A":0.0,"E":0.0,"F":0.0,"M":2.0,"L":0.0,"T":0.0}]},{"PRI_DEP":"14700000","Primary_Affiliation":[{"A":0.0,"E":3.0,"F":0.0,"M":1.0,"L":1.0,"T":19.0}]},{"PRI_DEP":"95011000","Primary_Affiliation":[{"A":0.0,"E":0.0,"F":1.0,"M":0.0,"L":0.0,"T":0.0}]},{"PRI_DEP":"Null","Primary_Affiliation":[{"A":0.0,"E":1.0,"F":0.0,"M":0.0,"L":0.0,"T":0.0}]},{"PRI_DEP":"ST010000","Primary_Affiliation":[{"A":1.0,"E":0.0,"F":0.0,"M":0.0,"L":0.0,"T":1.0}]}] 

LOA JSON:

[{"PRI_DEP":" ","LOA":[{"Blue":0.0,"UFM":0.0,"Invalid":0.0,"Bronze":1.0,"Basic":1.0}]},{"PRI_DEP":"14700000","LOA":[{"Blue":14.0,"UFM":5.0,"Invalid":1.0,"Bronze":4.0,"Basic":0.0}]},{"PRI_DEP":"95011000","LOA":[{"Blue":0.0,"UFM":0.0,"Invalid":0.0,"Bronze":1.0,"Basic":0.0}]},{"PRI_DEP":"Null","LOA":[{"Blue":0.0,"UFM":0.0,"Invalid":0.0,"Bronze":1.0,"Basic":0.0}]},{"PRI_DEP":"ST010000","LOA":[{"Blue":0.0,"UFM":0.0,"Invalid":1.0,"Bronze":0.0,"Basic":1.0}]}] 

状态JSON:

[{"PRI_DEP":" ","STATE":[{"A":2.0,"E":0.0}]},{"PRI_DEP":"14700000","STATE":[{"A":23.0,"E":1.0}]},{"PRI_DEP":"95011000","STATE":[{"A":1.0,"E":0.0}]},{"PRI_DEP":"Null","STATE":[{"A":1.0,"E":0.0}]},{"PRI_DEP":"ST010000","STATE":[{"A":2.0,"E":0.0}]}] 

现在，我想通过PRI_DEP以某种方式在一个JSON中表示这些内容

因此，所需的JSON如下所示（为便于阅读而更新）：

---

我一直在尝试不同的数据帧组合方法，我想我找到了答案

在我的原始帖子（设置嵌套组）中的python代码之后，我执行了以下操作：

Group_frames = [Nested_PAFF_df.set_index('PRI_DEP'), Nested_LOA_df.set_index('PRI_DEP'), Nested_ST_df.set_index('PRI_DEP')]
result = pd.concat(Group_frames, axis=1).reset_index()
print(result.to_json(orient='records'))

---

看来你想要的JSON被切断了。你们能更新吗？我故意只放第一条记录，但我会用剩下的记录更新。只有几个了。

[{"PRI_DEP":" ",
    "Primary_Affiliation":
        [{"A":0.0,"E":0.0,"F":0.0,"M":2.0,"L":0.0,"T":0.0}],
    "LOA": 
        [{"Blue":0.0,"UFM":0.0,"Invalid":0.0,"Bronze":1.0,"Basic":1.0}],
    "STATE":
        [{"A":2.0,"E":0.0}]},
 {"PRI_DEP":"14700000",
    "Primary_Affiliation": 
        [{"A":0.0,"E":3.0,"F":0.0,"M":1.0,"L":1.0,"T":19.0}],
    "LOA": 
        [{"Blue":14.0,"UFM":5.0,"Invalid":1.0,"Bronze":4.0,"Basic":0.0}],
    "STATE":
        [{"A":23.0,"E":1.0}]}, 
 {"PRI_DEP":"95011000",
    "Primary_Affiliation":
        [{"A":0.0,"E":0.0,"F":1.0,"M":0.0,"L":0.0,"T":0.0}],
    "LOA":
        [{"Blue":0.0,"UFM":0.0,"Invalid":0.0,"Bronze":1.0,"Basic":0.0}],
    "STATE":
        [{"A":1.0,"E":0.0}]},
 {"PRI_DEP":"Null",
    "Primary_Affiliation": 
        [{"A":0.0,"E":1.0,"F":0.0,"M":0.0,"L":0.0,"T":0.0}],
    "LOA":
        [{"Blue":0.0,"UFM":0.0,"Invalid":0.0,"Bronze":1.0,"Basic":0.0}],
    "STATE":
        [{"A":1.0,"E":0.0}]},
 {"PRI_DEP":"ST010000",
    "Primary_Affiliation":
        [{"A":1.0,"E":0.0,"F":0.0,"M":0.0,"L":0.0,"T":1.0}],
    "LOA":
        [{"Blue":0.0,"UFM":0.0,"Invalid":1.0,"Bronze":0.0,"Basic":1.0}],
    "STATE":
        [{"A":2.0,"E":0.0}]}]

Group_frames = [Nested_PAFF_df.set_index('PRI_DEP'), Nested_LOA_df.set_index('PRI_DEP'), Nested_ST_df.set_index('PRI_DEP')]
result = pd.concat(Group_frames, axis=1).reset_index()
print(result.to_json(orient='records'))