如何在Pandas中按MultiIndex与前置值交替替换数据？

Got it, let's work through this problem step by step. You want to keep the first column's value in each MultiIndex group (Set 1/Set 2), then alternate between 1 and 2 in pairs for the remaining columns—all based on that initial row value. Here's a clean, scalable way to make this happen:

Step 1: Define a Group Processing Function #

First, we'll create a function that takes a single column group (like all columns under "Set 1") and transforms each row to match your desired pattern:

def process_group(group):
    # Loop through each row in the current group
    for index, row in group.iterrows():
        first_val = row.iloc[0]
        # Set up the repeating pattern based on the first value of the row
        if first_val == 1:
            # Pattern: start with 1, then two 2s, two 1s, repeat
            base_pattern = [1, 2, 2, 1, 1]
        else:
            # Pattern: start with 2, then two 1s, two 2s, repeat
            base_pattern = [2, 1, 1, 2, 2]
        # Generate a sequence long enough to cover all columns in the group
        total_cols = len(row)
        repeated_pattern = (base_pattern * ((total_cols // len(base_pattern)) + 1))[:total_cols]
        # Replace the original row with our custom pattern
        group.loc[index] = repeated_pattern
    return group

Step 2: Apply the Function to Each MultiIndex Group #

We'll use pandas' groupby on the column MultiIndex's first level (the "Set X" labels) to apply our function to each group:

# Process each column group and get the transformed DataFrame
df_processed = df.groupby(level=0, axis=1).apply(process_group)

Verify the Result #

If you print df_processed, you'll see exactly the output you're looking for:

Quick Notes #

• This solution works for groups of any column length, not just the 5 and 8 columns in your example—it repeats the base pattern as needed to fit.
• If your data ever has initial values other than 1 or 2, you can easily add extra conditionals to handle those edge cases.

内容的提问来源于stack exchange，提问作者Herwini