Python 返回多索引数据帧中满足逻辑索引条件的每个组的最后一行

希望返回包含每个组的最后一行（具有最新日期索引的行）的数据帧，其中多索引的第二级由逻辑索引条件过滤

以下是一个玩具示例，用于更好地解释：

import numpy as np
import pandas as pd
from datetime import datetime

dates  = pd.date_range(start='1/1/2018', end='1/4/2018').to_pydatetime().tolist() * 2
ids    = ['z7321', 'z7321', 'z7321', 'z7321', 'b2134', 'b2134', 'b2134', 'b2134']

arrays = [ids, dates]
tuples = list(zip(*arrays))
index = pd.MultiIndex.from_tuples(tuples, names=['key', 'date'])

df = pd.DataFrame(data=np.random.randn(len(index)), index=index, columns=['change'])

print(df)

                   change
key   date
z7321 2018-01-01 -0.701605
      2018-01-02 -0.934580
      2018-01-03  0.186554
      2018-01-04  0.417024
b2134 2018-01-01  0.682699
      2018-01-02 -0.913633
      2018-01-03  0.330347
      2018-01-04 -0.706429

---

条件是返回最后一行，其中

df[df.index.get_level_values（1）。在我编写玩具示例时，我最终找到了一种获得所需输出的方法。希望此解决方案对其他人有帮助，或者可以改进

以下内容提供了所需的输出：

df1 = df[df.index.get_level_values(1) <= datetime(2018, 1, 2)].groupby(level='key', as_index=False).nth(-1)
print(df1)

                   change
key   date
z7321 2018-01-02 -0.934580
b2134 2018-01-02 -0.913633

df1=df[df.index.get_level_values（1）这是否回答了您的问题？另请参见：。是的，它确实有助于回答问题，我在搜索时没有遇到它。应用.tail（1）也可以。
df1 = df[df.index.get_level_values(1) <= datetime(2018, 1, 2)].groupby(level='key', as_index=False).nth(-1)
print(df1)

                   change
key   date
z7321 2018-01-02 -0.934580
b2134 2018-01-02 -0.913633

import numpy as np
import pandas as pd
from datetime import datetime

dates = pd.date_range(start='1/1/2018', end='1/4/2018').to_pydatetime().tolist()
dates += pd.date_range(start='12/29/2017', end='1/1/2018').to_pydatetime().tolist()

ids   = ['z7321', 'z7321', 'z7321', 'z7321', 'b2134', 'b2134', 'b2134', 'b2134']

arrays = [ids, dates]
tuples = list(zip(*arrays))
index = pd.MultiIndex.from_tuples(tuples, names=['key', 'date'])

df = pd.DataFrame(data=np.random.randn(len(index)), index=index, columns=['change'])
print(df)

                    change
key   date
z7321 2018-01-01 -1.420757
      2018-01-02 -0.297835
      2018-01-03  0.693520
      2018-01-04  0.909420
b2134 2017-12-29 -1.577685
      2017-12-30  0.632395
      2017-12-31  1.158273
      2018-01-01 -0.242314


df1 = df[df.index.get_level_values(1) <= datetime(2018, 1, 2)].groupby(level='key', as_index=False).nth(-1)
print(df1)

                    change
key   date
z7321 2018-01-02 -0.297835
b2134 2018-01-01 -0.242314