将Pandas日时间步长转换为每年52周（最后一周8天）的时间步长

Got it, let's work through this custom weekly averaging task together. Your requirement is clear: split each year into 52 weeks starting on January 1, with the first 51 weeks being 7 days each, and the final week covering the remaining 8 days (since 51*7 + 8 = 365, matching your 365-day year scope). Here's a step-by-step implementation with Pandas:

Step 1: Verify Datetime Index #

First, double-check that your DataFrame's index is properly formatted as datetime (it looks like it already is, but this is a quick sanity check):

import pandas as pd

# Ensure the index is datetime type
oldman.index = pd.to_datetime(oldman.index)

Step 2: Calculate Custom Week Numbers #

We need to map each date to its custom week within the year. We'll do this by:

1. Extracting the year for each date
2. Getting the day-of-year (1 to 365)
3. Assigning week numbers: days 1-357 go into weeks 1-51 (7 days each), days 358-365 go into week 52

# Add year and day-of-year columns
oldman['year'] = oldman.index.year
oldman['day_of_year'] = oldman.index.dayofyear

# Compute custom week number
oldman['custom_week'] = oldman['day_of_year'].apply(
    lambda x: (x - 1) // 7 + 1 if x <= 51*7 else 52
)

Let me break that lambda down:

• (x-1) //7 +1 converts day-of-year to a 1-based week number (e.g., days 1-7 become week 1, days 8-14 become week 2, etc.)
• Any day after 357 (51*7) gets assigned to week 52, which will always have exactly 8 days for 365-day years.

Step 3: Compute Weekly Averages #

Now group the data by year and custom week, then calculate the mean of the Value column:

# Group by year and custom week, calculate mean
weekly_averages = oldman.groupby(['year', 'custom_week'])['Value'].mean().reset_index()

# Optional: Format a readable week label (e.g., "1992-W01", "1992-W52")
weekly_averages['week_id'] = weekly_averages.apply(
    lambda row: f"{row['year']}-W{row['custom_week']:02d}",
    axis=1
)
weekly_averages.set_index('week_id', inplace=True)

Quick Validation #

To make sure this works, you can check the size of the 52nd week for any year:

# Check number of days in 1992's week 52
print(oldman[(oldman['year'] == 1992) & (oldman['custom_week'] == 52)].shape[0])
# Should output 8

Note on Leap Years #

If your dataset includes leap years (366 days), you'll need to decide how to handle the extra day. You could either extend week 52 to 9 days, or adjust the cutoff to 356 days (so week 51 has 7 days, week 52 has 9 days). The code above assumes 365-day years as per your original description.

内容的提问来源于stack exchange，提问作者kasra545