如何基于DataFrame列唯一值高效迭代行并筛选数据？

Let's fix that slow loop with a fully vectorized approach—this will handle your 60k rows in a fraction of the time, no per-product loops needed.

Core Solution Code #

import pandas as pd

# First, create a flag for each row indicating if its product has a 12-month entry
has_12_month = result.groupby('dpl_Description w/o months')['dpl_Order Duration'].transform(
    lambda x: (x == 12).any()
)

# Build the filter condition:
# - Keep all rows if the product doesn't have a 12-month entry, OR
# - If the product does have a 12-month entry, keep rows where Charity is "Yes" OR duration is 12/24/36
filter_condition = (
    ~has_12_month 
    | (result['Charity'] == "Yes") 
    | (result['dpl_Order Duration'].isin([12, 24, 36]))
)

# Apply the filter
result_filtered = result[filter_condition].reset_index(drop=True)

Why This Is Way Faster #

Your original loops are slow because:

• They repeatedly slice and modify the original DataFrame (inplace=True drops are expensive, especially in loops)
• Iterating over unique products and then individual rows/durations forces Pandas to work against its optimized vectorized design

This approach uses Pandas' built-in groupby.transform to create a flag for every row in one pass, then applies a single vectorized filter. All operations happen in bulk, leveraging Pandas' C-optimized backend instead of Python-level loops.

Verification with Your Sample Data #

For your sample dataset (assuming all Charity values are "No"), this code will:

• Keep all rows for products X, F, Z, C (none have a 12-month entry)
• For product Y (which has a 12-month entry), only keep the row with duration 12 (since other Y rows don't meet the duration criteria and aren't charity entries)
  This matches exactly your expected output.

Side Note: Better Ways to Share Tabular Data #

For Stack Overflow or similar platforms, here are cleaner options:

1. Markdown Tables: Use pd.DataFrame.to_markdown() to generate a ready-to-paste Markdown table:

   print(result.to_markdown(index=False))
   

   Which outputs a neatly formatted table like:

   dpl_Description w/o months	dpl_Order Duration	Charity
   X	36	No
   X	9	No
   Y	23	No

2. Reproducible Code Snippet: Share a list of lists/dictionaries that others can load directly into a DataFrame, avoiding manual formatting work.

内容的提问来源于stack exchange，提问作者ylux