Python 多指标不同顺序串联大Pandas 模型

我有两个数据帧，应该连接起来. 两者都是具有相同索引的多索引数据帧，但顺序不同.

所以，第一个数据帧的索引(df)看起来像:

MultiIndex([(11,  1, 1),
            (11,  1, 2),
            (11,  1, 3),
            ...
            (11, 24, 5),
            (11, 24, 6),
            (11, 24, 7)],
           names=['id_a', 'id_b', 'id_c'], length=168)

第二个看起来像:

MultiIndex([(11, 1,  1),
            (11, 2,  1),
            (11, 3,  1),
            (11, 3,  2),
            ...
            (11, 5, 23),
            (11, 6, 23),
            (11, 7, 23),
            (11, 7, 24)],
           names=['id_a', 'id_c', 'id_b'], length=168)

正如您所看到的，索引的顺序不同. 现在，通过运行pd.concat([df, df2]).index.names，我得到了以下结果:

FrozenList(['id_a', None, None])

如何重现

import pandas as pd

# create first data frame
idx = pd.MultiIndex.from_product(
    [['A1', 'A2', 'A3'], ['B1', 'B2', 'B3'], ['C1', 'C2', 'C3']],
    names=['a', 'b', 'c'])
cols = ['2010', '2020']
df = pd.DataFrame(1, idx, cols)

# Create second data frame with varying order
idx = pd.MultiIndex.from_product(
    [['A1', 'A2', 'A3'], ['C1', 'C2', 'C3'], ['B1', 'B2', 'B3']],
    names=['a', 'c', 'b'])
df2 = pd.DataFrame(2, idx, cols)

result = pd.concat([df, df2])

输出

> df
          2010  2020
a  b  c             
A1 B1 C1     1     1
      C2     1     1
      C3     1     1
   B2 C1     1     1
      C2     1     1
      C3     1     1
   B3 C1     1     1
      C2     1     1
      C3     1     1
A2 B1 C1     1     1
      C2     1     1
      C3     1     1
   B2 C1     1     1
      C2     1     1
      C3     1     1
   B3 C1     1     1
      C2     1     1
      C3     1     1
A3 B1 C1     1     1
      C2     1     1
      C3     1     1
   B2 C1     1     1
      C2     1     1
      C3     1     1
   B3 C1     1     1
      C2     1     1
      C3     1     1
> df2
          2010  2020
a  c  b             
A1 C1 B1     2     2
      B2     2     2
      B3     2     2
   C2 B1     2     2
      B2     2     2
      B3     2     2
   C3 B1     2     2
      B2     2     2
      B3     2     2
A2 C1 B1     2     2
      B2     2     2
      B3     2     2
   C2 B1     2     2
      B2     2     2
      B3     2     2
   C3 B1     2     2
      B2     2     2
      B3     2     2
A3 C1 B1     2     2
      B2     2     2
      B3     2     2
   C2 B1     2     2
      B2     2     2
      B3     2     2
   C3 B1     2     2
      B2     2     2
      B3     2     2
> result
          2010  2020
a                   
A1 B1 C1     1     1
      C2     1     1
      C3     1     1
   B2 C1     1     1
      C2     1     1
      C3     1     1
   B3 C1     1     1
      C2     1     1
      C3     1     1
A2 B1 C1     1     1
      C2     1     1
      C3     1     1
   B2 C1     1     1
      C2     1     1
      C3     1     1
   B3 C1     1     1
      C2     1     1
      C3     1     1
A3 B1 C1     1     1
      C2     1     1
      C3     1     1
   B2 C1     1     1
      C2     1     1
      C3     1     1
   B3 C1     1     1
      C2     1     1
      C3     1     1
A1 C1 B1     2     2
      B2     2     2
      B3     2     2
   C2 B1     2     2
      B2     2     2
      B3     2     2
   C3 B1     2     2
      B2     2     2
      B3     2     2
A2 C1 B1     2     2
      B2     2     2
      B3     2     2
   C2 B1     2     2
      B2     2     2
      B3     2     2
   C3 B1     2     2
      B2     2     2
      B3     2     2
A3 C1 B1     2     2
      B2     2     2
      B3     2     2
   C2 B1     2     2
      B2     2     2
      B3     2     2
   C3 B1     2     2
      B2     2     2
      B3     2     2
> result.index.names
FrozenList(['a', None, None])

索引"b"和"c"消失了.