Pandas行内多值校验与替换问题求助：现有代码未生效

Hey there! Let's sort out this pandas DataFrame problem together. Your original code didn't work for a couple of small but important reasons—let's break it down and fix it with clean, standard code that meets both your requirements.

Step 1: Set Up Example Data #

First, let's recreate your DataFrame (plus an extra all-0 row to test the second requirement):

import pandas as pd

data = [
    [0, 10, 0, 0],
    [1, 1, 1, 1],
    [0, 12, 0, 0],
    [0, 0, 0, 0],  # Test case for requirement 2
    [0, 13, 0, 0]
]
df = pd.DataFrame(data, columns=['open', 'high', 'low', 'close'])

Step 2: Implement Requirement 1 #

When open, low, close are all 0 and high is not 0, replace those three columns with the high value:

# Define the exact condition for requirement 1
condition1 = (df['open'] == 0) & (df['low'] == 0) & (df['close'] == 0) & (df['high'] != 0)

# Replace the columns—note the list for column selection and reshaping to match dimensions
df.loc[condition1, ['open', 'low', 'close']] = df.loc[condition1, 'high'].values.reshape(-1, 1)

Step 3: Implement Requirement 2 #

When all four columns are 0, set open/low to 1 and close/high to 10:

# Define the condition for all values being 0 (shorter than checking each column individually)
condition2 = (df == 0).all(axis=1)

# Assign the new values to the respective columns
df.loc[condition2, ['open', 'low']] = 1
df.loc[condition2, ['close', 'high']] = 10

Step 4: Check the Result #

Run print(df) and you'll get the corrected DataFrame:

open  high  low  close
0    10    10   10     10
1     1     1    1      1
2    12    12   12     12
3     1    10    1     10
4    13    13   13     13

Why Your Original Code Failed #

• Incomplete conditions: You only checked open == 0, but didn't verify low/close were also 0 or exclude rows where high was 0. This would have incorrectly modified all-0 rows too.
• Column selection syntax: In loc, multiple columns need to be passed as a list (['open', 'low', 'close']) instead of comma-separated names.
• Dimension mismatch: Assigning a 1-dimensional Series directly to multiple columns causes a shape error—using reshape(-1,1) converts it to a 2D array that matches the column structure.

内容的提问来源于stack exchange，提问作者Marx Babu