Python pandas 创建不均匀的多索引

Question

我有以下代码，

IDX_VALS_BANKNOTER_PATRIMONY = [['PATRIMONY'],['GOLD']]
IDX_VALS_BANKNOTER_ASSETS = [['ASSETS'],['DEPOSITS', 'ADVANCES']]
IDX_VALS_BANKNOTER_LIABILITIES = [['LIABILITIES'], ['CLIENTS', 'SUPPLIERS']]

IDX_BANKNOTER_PATRIMONY = pd.MultiIndex.from_product(IDX_VALS_BANKNOTER_PATRIMONY)
IDX_BANKNOTER_ASSETS = pd.MultiIndex.from_product(IDX_VALS_BANKNOTER_ASSETS)
IDX_BANKNOTER_LIABILITIES = pd.MultiIndex.from_product(IDX_VALS_BANKNOTER_LIABILITIES)

IDX_BANKNOTER = IDX_BANKNOTER_PATRIMONY.append(IDX_BANKNOTER_ASSETS).append(IDX_BANKNOTER_LIABILITIES)

print(IDX_BANKNOTER)

打印以下索引：

MultiIndex([(  'PATRIMONY',      'GOLD'),
            (     'ASSETS',  'DEPOSITS'),
            (     'ASSETS',  'ADVANCES'),
            ('LIABILITIES',   'CLIENTS'),
            ('LIABILITIES', 'SUPPLIERS')],
           )

（我使用.from_product()是因为我希望最终添加更多标签）我的问题如下：我想在第三列上扩展这个多索引，以便我得到一个如下所示的多索引：

'PATRIMONY', 'GOLD',
'ASSETS', 'DEPOSITS',
'ASSETS', 'ADVANCES',
'LIABILITIES', 'CLIENTS', 'Dr. Foo'
'LIABILITIES', 'CLIENTS', 'Dr. House'
'LIABILITIES', 'CLIENTS', 'Richard'
'LIABILITIES', 'SUPPLIERS', 'PORT1',
'LIABILITIES', 'SUPPLIERS', 'PORT2'

这意味着多重索引将是不均匀的，第三个级别仅由“LIABILITIES”使用，并且根据客户名称或供应商名称为 CLIENTS 和 SUPPLIERS 使用不同的索引。我尝试附加以下索引：

IDX_FIRST_EXTENSION_NAMES = [['LIABILITIES'], ['CLIENTS'], ['Dr. Foo', 'Dr. House', 'Richard']]
IDX_FIRST_EXTENSION = pd.MultiIndex.from_product(IDX_FIRST_EXTENSION_NAMES)
IDX_SECOND_EXTENSION_NAMES = [['LIABILITIES'], ['SUPPLIERS'], ['PORT1', 'PORT2']]
IDX_SECOND_EXTENSION = pd.MultiIndex.from_product(IDX_SECOND_EXTENSION_NAMES)
DESIRED_RESULT = IDX_BANKNOTER.append(IDX_FIRST_EXTENSION).append(IDX_SECOND_EXTENSION)

但我得到的回报是：

MultiIndex([(  'PATRIMONY',      'GOLD'),
            (     'ASSETS',  'DEPOSITS'),
            (     'ASSETS',  'ADVANCES'),
            ('LIABILITIES',   'CLIENTS'),
            ('LIABILITIES',   'CLIENTS'),
            ('LIABILITIES',   'CLIENTS'),
            ('LIABILITIES', 'SUPPLIERS'),
            ('LIABILITIES', 'SUPPLIERS')],
           )

我对使用熊猫相当陌生，关于多索引的文档并没有帮助（它有相当数量的初始化多索引的示例，并且没有不均匀多索引的示例）。有人有指点吗？我正在制作这个多索引以方便操作相应的数据，例如能够访问特定的客户帐户

df['LIABILITIES']['CLIENTS']['(CLIENT NAME)']

或者能够得到下所有值的总和['CLIENTS']。理想情况下，我希望保留数据帧的列作为时间标签。

任何帮助表示赞赏，谢谢。

Answer 1

代码：

import pandas as pd

IDX_VALS_BANKNOTER_PATRIMONY = [['PATRIMONY'],['GOLD'], ['']]
IDX_VALS_BANKNOTER_ASSETS = [['ASSETS'],['DEPOSITS', 'ADVANCES'], ['']]

IDX_BANKNOTER_PATRIMONY = pd.MultiIndex.from_product(IDX_VALS_BANKNOTER_PATRIMONY)
IDX_BANKNOTER_ASSETS = pd.MultiIndex.from_product(IDX_VALS_BANKNOTER_ASSETS)

IDX_BANKNOTER = IDX_BANKNOTER_PATRIMONY.append(IDX_BANKNOTER_ASSETS)

IDX_FIRST_EXTENSION_NAMES = [['LIABILITIES'], ['CLIENTS'], ['Dr. Foo', 'Dr. House', 'Richard']]
IDX_FIRST_EXTENSION = pd.MultiIndex.from_product(IDX_FIRST_EXTENSION_NAMES)
IDX_SECOND_EXTENSION_NAMES = [['LIABILITIES'], ['SUPPLIERS'], ['PORT1', 'PORT2']]
IDX_SECOND_EXTENSION = pd.MultiIndex.from_product(IDX_SECOND_EXTENSION_NAMES)
WANTED_RESULT = IDX_BANKNOTER.append(IDX_FIRST_EXTENSION).append(IDX_SECOND_EXTENSION)

print(WANTED_RESULT)

输出：

MultiIndex([(  'PATRIMONY',      'GOLD',          ''),
            (     'ASSETS',  'DEPOSITS',          ''),
            (     'ASSETS',  'ADVANCES',          ''),
            ('LIABILITIES',   'CLIENTS',   'Dr. Foo'),
            ('LIABILITIES',   'CLIENTS', 'Dr. House'),
            ('LIABILITIES',   'CLIENTS',   'Richard'),
            ('LIABILITIES', 'SUPPLIERS',     'PORT1'),
            ('LIABILITIES', 'SUPPLIERS',     'PORT2')],
           )