如何使用mask（掩码）合并Phone 1与Phone 2两列数据？

Got it! Since you already have that 0/1 mask column, merging your Phone1 and Phone2 columns into a single phone column is totally straightforward. We can use the mask to conditionally pick values from either column—let's use pandas (the go-to tool for tabular data) to walk through this with your simulated data.

Step 1: Set up sample data #

First, let's replicate your scenario with some mock data:

import pandas as pd

# Simulate your dataset
df = pd.DataFrame({
    'Phone1': ['123-456-7890', None, '987-654-3210', None],
    'Phone2': [None, '555-1234', None, '555-4321'],
    'mask': [1, 0, 1, 0]  # Assume mask=1 means use Phone1, mask=0 means use Phone2
})

Step 2: Merge columns with the mask #

You have two simple, clean ways to do this:

Option 1: Use numpy.where

This is super intuitive—it works like an inline if-else:

import numpy as np

# When mask is 1, take Phone1; else take Phone2
df['phone'] = np.where(df['mask'] == 1, df['Phone1'], df['Phone2'])

Option 2: Use pandas' built-in where method

Pandas has its own where function that does the same thing, with a slightly different syntax:

# Keep Phone1 where mask is 1, otherwise replace with Phone2
df['phone'] = df['Phone1'].where(df['mask'] == 1, df['Phone2'])

Step 3: Check the result #

Running either of the above will give you your desired phone column:

print(df)
# Output:
        Phone1     Phone2  mask         phone
0  123-456-7890       None     1  123-456-7890
1          None  555-1234     0      555-1234
2  987-654-3210       None     1  987-654-3210
3          None  555-4321     0      555-4321

Bonus: Handle edge cases (optional) #

If you want to fall back to the other phone number when your selected column is empty, you can add a fillna layer:

# Use mask to pick primary column, then fill empty values with the other column
df['phone'] = np.where(
    df['mask'] == 1,
    df['Phone1'].fillna(df['Phone2']),
    df['Phone2'].fillna(df['Phone1'])
)

Just adjust the mask condition (e.g., df['mask'] == 0) if your mask logic is reversed (0 means use Phone1 instead of 1).

内容的提问来源于stack exchange，提问作者Yun Tae Hwang