Python中如何保存含自定义列名的coef_matrix_simple至文本文件？

Got it, let's sort out how to save your coef_matrix_simple DataFrame along with its column names (col) to a text file. The problem with your original np.savetxt() approach is that it’s built for raw numeric arrays—so it doesn’t automatically preserve column headers or row indices by default.

Simplest Solution: Use Pandas' to_csv() #

Pandas' DataFrame has a built-in to_csv() method that handles headers and indices out of the box, which is way more straightforward than numpy's method for this use case. Here's your modified code:

col = ['rss','intercept'] + ['coef_x_%d'%i for i in range(1,16)]
ind = ['model_pow_%d'%i for i in range(1,16)]
coef_matrix_simple = pd.DataFrame(index=ind, columns=col)
for i in range(1,16):
    coef_matrix_simple.iloc[i-1,0:i+2] = linear_regression(df, power=i, models_to_plot=models_to_plot)
pd.options.display.float_format = '{:,.2g}'.format

# Save with column headers and row indices (matches your original space delimiter)
coef_matrix_simple.to_csv('output.csv', sep=' ', float_format='{:,.2g}'.format)

Key Details:

• sep=' ': Keeps the space delimiter you used with np.savetxt()
• float_format='{:,.2g}'.format: Ensures the number formatting matches what you set with pd.options.display.float_format
• If you don't want to save the row indices, add index=False to the to_csv() call:

  coef_matrix_simple.to_csv('output.csv', sep=' ', float_format='{:,.2g}'.format, index=False)
  

Alternative: Using np.savetxt() (If You Must) #

If you need to stick with numpy for some reason, you’ll have to manually write the header and combine indices with your data. This is more clunky, but here’s how:

col = ['rss','intercept'] + ['coef_x_%d'%i for i in range(1,16)]
ind = ['model_pow_%d'%i for i in range(1,16)]
coef_matrix_simple = pd.DataFrame(index=ind, columns=col)
for i in range(1,16):
    coef_matrix_simple.iloc[i-1,0:i+2] = linear_regression(df, power=i, models_to_plot=models_to_plot)
pd.options.display.float_format = '{:,.2g}'.format

# Manually write header and data
with open('output.csv', 'w') as f:
    # Write column headers as a space-separated line
    f.write(' '.join(col) + '\n')
    # Combine row indices with numeric data
    data_with_indices = np.hstack([np.array(coef_matrix_simple.index).reshape(-1, 1), coef_matrix_simple.values])
    # Save the combined data
    np.savetxt(f, data_with_indices, delimiter=' ', fmt='%s')

This works, but you have to handle formatting for both string indices and numeric values, which is why the pandas method is preferred.

内容的提问来源于stack exchange，提问作者SANDIPAN DAWN