将 pandas dataframe 转换为嵌套字典/json 格式

Question

我正在尝试将以下 pandas 数据帧（python）转换为嵌套字典格式。 输入数据 pandas dataframe ：

statetraffic   |state    | act       | traffic| reward | header      | time |   id

  stateacttraf |     1   |    1      | 12      |  22     |   str1    |   1572340221000 | 34022100
  stateacttraf |     1   |    2      | 87      |  30     |   str1    |   1572340221000 | 34022100
  stateacttraf |     1   |    3      | 1       |  48     |   str1    |   1572340221000 | 34022100
  stateacttraf |     2   |    1      | 10      |  13     |   str1    |   1572340221000 | 34022100
  stateacttraf |     2   |    2      | 80      |  27     |   str1    |   1572340221000 | 34022100
  stateacttraf |     2   |    3      | 10      |  60     |   str1    |   1572340221000 | 34022100

已尝试以下代码但没有奏效：

1)final_op = input_df.to_dict(orient='records') -> does not provide the answer       
2)from jsonmerge import merge; 
message = {'statetraffic': 'stateacttraf'}; 
message1 = {'time': time.time()}; 
result = merge(final_op, message, message2) -> Neither does this provide the answer either

需要某种形式的嵌套字典

期望字典/json output像这样：

{

{  "statetraffic":"stateacttraf",
   "time":1572340221000,
   "str1":{ 
      "id":34022100,
      "state":1,
      "act":1,
      "trafficSplit":12,
      "reward":22
   }
{ 
   "statetraffic":"stateacttraf",
   "time":1572340221000,
   "str1":{ 
      "id":34022100,
      "state":1,
      "act":2,
      "trafficSplit":87,
      "reward":30
   }
{ 
   "statetraffic":"stateacttraf",
   "time":1572340221000,
   "str1":{ 
      "id":34022100,
      "state":1,
      "act":3,
      "trafficSplit":1,
      "reward":48
   }
{  "statetraffic":"stateacttraf",
   "time":1572340221000,
   "str1":{ 
      "id":34022100,
      "state":2,
      "act":1,
      "trafficSplit":10,
      "reward":13
   }
}

迫切需要这种格式的 output。 因此，任何帮助将不胜感激。

Answer 1

试试这个，假设你的 dataframe 为df

main_dict = df.to_dict()
uprow= ["statetraffic","time","header"]
drow = ["id","state" ,"act" ,"traffic","reward"]

datalist = []
for c in range(df.shape[0]):
    subd = {}
    for k,v in main_dict.items():
        subd[k] = v[c]

    subd_ = subd.copy()
    tmp = subd.get("header")

    subd[tmp] = 0
    for i in uprow: del subd_[i]
    subd[tmp]=subd_
    for i in drow: del subd[i] 
    del subd["header"]
    datalist.append(subd)

print(datalist)