Jupyter Notebook中Pandas DataFrame复制后原变量被意外修改的原因及冻结方法问询

Hey there, this isn't an environment bug—it's a classic Python/Pandas gotcha with mutable objects!

When you write temp_df = original_df, you're not making a new copy of your DataFrame. Instead, you're just giving the same underlying data a second name. So any changes you make to temp_df directly edit that shared data, which is why original_df ends up looking identical. This is how Python handles mutable objects (like DataFrames, lists, dictionaries) by default—assignments are references, not copies.

The fix is straightforward: use Pandas' .copy() method to create a separate, independent version of your DataFrame. This way, you can modify temp_df without touching the original, and you don't have to re-run the initial cell that creates original_df.

Here's what you need to do instead of temp_df = original_df:

• For a full, independent copy (this is almost always what you want to ensure original_df stays untouched):

  temp_df = original_df.copy(deep=True)
  

• If you only need a shallow copy (copies the DataFrame structure but shares nested mutable elements like lists in cells):

  temp_df = original_df.copy()  # Shallow is the default, so you can omit `deep=False`
  

Using deep=True ensures every part of the DataFrame is duplicated, so original_df stays exactly as you created it—you can keep your Jupyter Notebook split into logical, step-by-step cells for visualization and analysis as intended.

Let's walk through a simple example to confirm:

1. Create your original DataFrame:

   import pandas as pd
   original_df = pd.DataFrame({'A': [1, 2, 3], 'B': [4, 5, 6]})
   

2. Make a proper copy:

   temp_df = original_df.copy(deep=True)
   

3. Modify temp_df:

   temp_df['A'] = [10, 20, 30]
   

4. Check original_df—it's still intact!

   print(original_df)
   # Output:
   #    A  B
   # 0  1  4
   # 1  2  5
   # 2  3  6
   

This solves your problem completely while preserving the step-by-step workflow that makes Jupyter Notebooks so useful.

内容的提问来源于stack exchange，提问作者Christian