Python Pandas：利用groupby标记各周期内的POI（最值点）

Your original code has a critical mistake: you're calling Data3.groupby('Cycle') where Data3 is the raw dictionary, not the pandas DataFrame df. Groupby operations must be performed on the DataFrame object, not the source dictionary.

Corrected Apply + np.select Approach #

This fixes your original approach by targeting the DataFrame for grouping, then aligning the result to the original index:

import pandas as pd
import numpy as np

Data3 = {'Cycle': ['1', '1', '1', '1', '1', '1', '2','2', '2', '2', '2', '2'],
        'Value': [20, 24, 25, 18,15,12,1,2,19,18,12,1],
        'Diff2':[1,10,18,-12,-50,-14,14,150,130,-140,12,14],
        } 

df = pd.DataFrame(Data3)

df['POI'] = df.groupby('Cycle').apply(
    lambda g: np.select([g['Value'] == g['Value'].max(),
                         g['Diff2'] == g['Diff2'].min()], ['A', 'B'], default='-')
).explode().reset_index(drop=True)

print(df)

More Efficient Transform Approach #

Using transform is cleaner and faster, as it directly returns values aligned with the original DataFrame index, eliminating the need for explode and index adjustment:

import pandas as pd
import numpy as np

Data3 = {'Cycle': ['1', '1', '1', '1', '1', '1', '2','2', '2', '2', '2', '2'],
        'Value': [20, 24, 25, 18,15,12,1,2,19,18,12,1],
        'Diff2':[1,10,18,-12,-50,-14,14,150,130,-140,12,14],
        } 

df = pd.DataFrame(Data3)

# Calculate group-wise max Value and min Diff2, aligned to original rows
max_value_per_cycle = df.groupby('Cycle')['Value'].transform('max')
min_diff2_per_cycle = df.groupby('Cycle')['Diff2'].transform('min')

# Apply conditions to set POI
df['POI'] = np.select(
    [df['Value'] == max_value_per_cycle, df['Diff2'] == min_diff2_per_cycle],
    ['A', 'B'],
    default='-'
)

print(df)

Output #

Both approaches produce the target DataFrame:

Cycle  Value  Diff2 POI
0      1     20      1   -
1      1     24     10   -
2      1     25     18   A
3      1     18    -12   -
4      1     15    -50   B
5      1     12    -14   -
6      2      1     14   -
7      2      2    150   -
8      2     19    130   A
9      2     18   -140   B
10     2     12     12   -
11     2      1     14   -

内容的提问来源于stack exchange，提问作者SeanK22