Pandas多列值按条件修改报错(KeyError)求助

Got it, let's break down why you're hitting that KeyError: ('column1', 'column2') and get your multi-column conditional update working properly.

Why Your Original Code Fails #

Two main issues are tripping you up:

1. Wrong column indexing syntax: When selecting multiple columns, you need to use a list (['column1', 'column2']) instead of a tuple (('column1', 'column2')). Writing df['column1','column2'] tells pandas to look for a single column named ('column1', 'column2')—which doesn't exist, hence the KeyError.
2. Missing row-wise condition check: Comparing a multi-column DataFrame to 1 gives a boolean DataFrame (each cell is True/False). You need to collapse this to a single boolean per row (to confirm both columns meet the condition) using .all(axis=1).

Correct Approach #

Here's how to properly update both columns where both column1 and column2 are >= 1:

First, let's create a sample DataFrame to test with:

import pandas as pd

# Sample data to demonstrate
df = pd.DataFrame({
    'column1': [0.8, 1.5, 2.2, 0.9],
    'column2': [1.1, 0.7, 1.8, 2.1]
})

Now apply the conditional update correctly:

# Step 1: Create a boolean mask for rows where both columns meet the condition
mask = df[['column1', 'column2']].ge(1).all(axis=1)

# Step 2: Use the mask to update the target columns
df.loc[mask, ['column1', 'column2']] = 1

Or combine it into a single concise line if you prefer:

df.loc[df[['column1', 'column2']].ge(1).all(axis=1), ['column1', 'column2']] = 1

What This Does #

• df[['column1', 'column2']].ge(1): Generates a boolean DataFrame where each cell is True if the value is >=1.
• .all(axis=1): Checks if all values in a row are True (meaning both columns meet the condition), returning a boolean Series we can use as a row filter.
• df.loc[mask, ['column1', 'column2']] = 1: Updates only the rows that pass the mask check, setting both target columns to 1.

Alternative (Less Efficient) Method #

If you're curious, you could use apply, but this is slower for large datasets since it operates row-by-row:

df[['column1', 'column2']] = df[['column1', 'column2']].apply(
    lambda x: [1, 1] if (x['column1'] >=1 and x['column2'] >=1) else x,
    axis=1,
    result_type='expand'
)

Stick with the vectorized method above—it's the pandas-idiomatic way to handle this, and way faster for big data.

内容的提问来源于stack exchange，提问作者Mostafa Alishahi