在pivot_表函数中对数据重新排序

Question

我有一个示例数据帧

region   headquarter      sales    base    %growth    month_year
   X      Los Angeles     1000     2000      30        202101
   X      Florida City    2000     2000      20        202101
   X      Los Angeles     5000     6000      70        202001
   X      Florida City    4000     4500      45        202001

我正在尝试使用

data = data.pivot_table(columns=['month_year'], values=['sales', 'base', '%growth'], index=['region', 'headquarter'])

print(data)

>                              %growth          base           sales
         |               |   202001  202101  202001 202101   202001 202101
   -----------------------------------------------------------------------
  region |  headquarter  |
     x   |  Los Angeles  |    70    30     6000   2000       5000    1000
         |  Florida City |    45    20     4000   2000       4500    2000

这些值的顺序与我在上面的代码片段中提到的不同

如何将数据重新构造为（通过重复行标签）

>                            202001                 202101
                         Sales  base  %growth    Sales  base  %growth     
   region   headquarter
     X      Los Angeles   5000   6000    70       1000   2000     30
     X      Florida City  4500   4000    45       2000   2000     20

Answer 1

将^{}与^{}一起使用：

mux = pd.MultiIndex.from_product([data['month_year'].unique(), ['Sales','base','%growth']])
data = data.pivot_table(columns=['month_year'], 
                        values=['Sales', 'base', '%growth'], 
                        index=['headquarter']).swaplevel(1, 0, axis=1).reindex(mux, axis=1)

print(data)
             202101               202001              
              Sales  base %growth  Sales  base %growth
headquarter                                           
Florida City   2000  2000      20   4000  4500      45
Los Angeles    1000  2000      30   5000  6000      70

编辑：

mux = pd.MultiIndex.from_product([data['month_year'].unique(), ['sales','base','%growth']])


data = data.pivot_table(columns=['month_year'], 
                        values=['sales', 'base', '%growth'], 
                        index=['region', 'headquarter']).swaplevel(1, 0, axis=1).reindex(mux, axis=1)


print (data)
                    202101               202001              
                     sales  base %growth  sales  base %growth
region headquarter                                           
X      Florida City   2000  2000      20   4000  4500      45
       Los Angeles    1000  2000      30   5000  6000      70