Question

我有下面的json文件，我想根据更新后的数据重新创建。

{"AAL": {"year": [2012, 2013, 2014, 2015], 
         "eps": [-5.6, -11.25, 4.02, 11.39], 
         "revenue": [24855.0, 26743.0, 42650.0, 40990.0], 
         "op_revenue": [148.0, 1399.0, 4249.0, 6204.0]}, 
 "AAP": {"year": [2012, 2013, 2014, 2015], 
         "eps": [5.29, 5.36, 6.75, 6.45], 
         "revenue": [6205.003000000001, 6493.814, 9843.860999999999, 9737.018], 
         "op_revenue": [657.315, 660.318, 851.71, 825.78]}, 
 "AAPL": {"year": [2013, 2014, 2015, 2016], 
          "eps": [40.03, 6.49, 9.28, 8.35], 
          "revenue": [171000.0, 183000.0, 234000.0, 216000.0], 
          "op_revenue": [48999.0, 52503.0, 71230.0, 60024.0]}
 ...}

我的数据来自形状完全相同的三个表格（ eps，收入，op_revenue ）。下面是一个表的前几行（第一列标题是股票行情指示器，其余列是年份）。

  ticker      2012        2013        2014       2015      2016       2017       2018
1      A 938000000   724000000   740000000  713000000 692000000  504000000  381000000
2    AAL 431000000 -1833000000 -1012000000 -752000000 -99000000 2499000000 2951000000
3    AAN 134624000   120666000   108005000   90656000  78813000   78233000   89137000
4   AAOI    390000     -131000      -46000    1873000   3060000    4283000    3523000
5   AAON  37359000    37547000    40229000   39473000  41391000   44158000   42735000
6    AAP 407546000   391758000   417694000  440311000 458658000  493825000  494211000

如何重新创建json文件？

Answer 1

考虑将每个数据帧从宽到长分解（即，不使用 year 数据值作为元素），然后与pandas.concat()串联，最后在< em> ticker + to_dict：

groupby

输出 （重复所有三个数据集的OP发布数据）

df_dict = {'eps': eps, 'revenue': revenue, 'op_revenue': op_revenue}

# MELTING WIDE TO LONG
new_df_dict = {k:(pd.melt(v, id_vars = "ticker", var_name = "year", value_name = k)
                    .set_index(["ticker", "year"])
                 ) for k,v in df_dict.items()}

# HORIZONTAL CONCATENATING
final_df = (pd.concat(new_df_dict, axis="columns")
              .sort_index()
              .reset_index()
           )

final_df.columns = final_df.columns.get_level_values(0)

# TICKER GROUPBY DICTIONARY
final_dict = {i: g.drop(columns='ticker').to_dict(orient='list') \
                for i,g in final_df.groupby('ticker')}

# OUTPUT TO JSON
with open('Output.json', 'w') as f:
    f.write(json.dumps(final_dict, indent=3))

从多个数据帧中提取数据

1 个答案: