DataFrame分组后Timedelta列求均值报错求助

Let's work through your problem step by step, addressing both the empty output and the aggregation error:

1. Why is your output length 0? #

This almost always means your intermediate time_to_rent DataFrame has no rows. Let's trace back to the source:

• First, validate if your initial filter returns any data:

  filtered_payments = user_payments[user_payments.rentComplete]
  print(filtered_payments.shape)  # Should show (number_of_rows, column_count)
  

  If this prints (0, X), your filter isn't capturing any data. Double-check the rentComplete column:
  • Is it a boolean column? If it uses string values (like 'Yes'/'No' or 'True'/'False'), adjust your filter to match, e.g., user_payments[user_payments.rentComplete == 'True'].
  • It’s also possible there are no rows where rentComplete is truthy at all—you’ll need to verify your raw dataset here.

2. Fixing the DataError: No numeric types to aggregate #

The error occurs because rent_time is a timedelta64 type, and np.mean doesn’t handle this type as seamlessly as pandas’ native aggregation tools. Here are two reliable fixes:

Option 1: Use pandas' built-in mean instead of np.mean #

Replace your aggregation line with this:

average_per_user = time_to_rent.groupby('creditCardId').agg({'rent_time': 'mean'})

Pandas natively supports averaging timedelta values, so this will return a timedelta result (e.g., 0 days 02:30:00 for an average of 2.5 hours).

Option 2: Convert timedelta to numeric seconds first #

If you prefer working with numeric values, convert rent_time to total seconds before aggregating:

# Convert timedelta to a numeric column (total seconds)
time_to_rent['rent_time_seconds'] = time_to_rent['rent_time'].dt.total_seconds()

# Now np.mean works perfectly with the numeric type
average_per_user = time_to_rent.groupby('creditCardId').agg({'rent_time_seconds': np.mean})

You can convert the result back to a timedelta later if needed:

average_per_user['average_rent_time'] = pd.to_timedelta(average_per_user['rent_time_seconds'], unit='s')

Final Checklist #

1. Confirm your filtered user_payments has rows before proceeding to groupby operations.
2. Use one of the aggregation fixes above to avoid the DataError.

内容的提问来源于stack exchange，提问作者OcMaRUS