如何简化N个同结构数据文件的对应元素求平均操作？

Absolutely! You can streamline this workflow to handle any number of columns without rewriting code every time a new variable is added. Here's a cleaner, more maintainable approach using numpy's flexible array handling:

Improved Code #

import numpy as np
import glob

# Grab all target text files
list_of_files = glob.glob('*.txt')

# Load all data into a single 3D array: (number_of_files, number_of_rows, number_of_columns)
# This automatically adapts to any number of columns in your files
all_data = np.array([np.loadtxt(f) for f in list_of_files])

# Calculate the mean across all files (average along the first axis)
mean_data = all_data.mean(axis=0)

# Preserve the original header from the first file
with open(list_of_files[0], 'r') as first_file:
    header = first_file.readline().strip()

# Write the mean results to output
with open("MeanFile.txt", 'w') as out_file:
    out_file.write(f"{header}\n")
    # Use numpy's savetxt for efficient, formatted writing
    np.savetxt(out_file, mean_data, fmt='%.3f')

Key Advantages Over Your Original Approach #

• No hardcoded column variables: You don't need to define separate arrays for p, q, s, etc. The 3D array all_data automatically captures all columns from your files, so adding a new column (like theta) requires zero code changes.
• Simplified averaging: Instead of calling .mean(0) on each individual array, a single mean(axis=0) computes the average across all files for every position in the dataset.
• Automatic header preservation: The code reads the header from the first file, ensuring your output file retains the original column labels without manual updates.
• Cleaner file writing: np.savetxt replaces manual loop-based writing, reducing the chance of formatting errors and making the code more efficient.

Notes #

• This assumes all your data files have identical structure (same number of rows, columns, and header format) — which matches your description. If you ever need to handle inconsistent files, you could add a quick check step to validate dimensions before loading.
• For extremely large datasets, loading all files into memory at once might be resource-heavy. In that case, you could load files one at a time and accumulate the sum, then divide by the number of files at the end (but this is rarely necessary for standard-sized datasets).

内容的提问来源于stack exchange，提问作者Francesco Di Lauro