Consider the following example

import numpy as np
import pandas as pd

data = {
    'Group': ['A', 'B', 'C', 'D']*3,  # Repeating groups to fill the DataFrame
    'Timestamp': pd.date_range(start='2023-01-01', periods=12, freq='M'),  # Monthly frequency
    'Numeric': np.random.rand(12) * 100,  # Random numeric values scaled up
    'String': ['apple', 'banana', 'cherry', 'date']*3  # Repeating string entries
}

df = pd.DataFrame(data)
df.head()
Out[28]: 
  Group  Timestamp    Numeric  String
0     A 2023-01-31  69.320654   apple
1     B 2023-02-28   1.667633  banana
2     C 2023-03-31  14.211651  cherry
3     D 2023-04-30  40.061005    date
4     A 2023-05-31  23.433903   apple

I know I can use chaining using pipe, which works really well. However, the following line of code, which tries to filter the groupby dataframe using .loc fails. What is the issue here? Isn’t pipe just passing a dataframe (so .loc should work)?

(df.groupby('Group')
 .pipe(lambda x: x.loc[x.Numeric <4]))

Traceback (most recent call last):
AttributeError: 'DataFrameGroupBy' object has no attribute 'loc'