Python中如何将DataFrame列中字符串转为NaN？指定特定值转NaN方法

Hey there! Let's break down your two Pandas questions with clear, actionable examples:

If you want to turn all non-numeric values (i.e., strings) in a column into NaN, the easiest way is to use pd.to_numeric() with the errors='coerce' parameter. This function tries to convert the column to a numeric type—any value that can't be converted (like strings) gets replaced with NaN automatically.

Example code:

import pandas as pd

# Sample DataFrame with mixed numeric and string values
df = pd.DataFrame({
    'value': [10, 'twenty', 30, 'forty']
})

# Convert all strings in 'value' to NaN
df['value'] = pd.to_numeric(df['value'], errors='coerce')

print(df)

Output:

value
0   10.0
1    NaN
2   30.0
3    NaN

For your specific case where you want to replace only the string 'B' in the 'type' column with NaN, here are two straightforward, reliable methods:

Method 1: Using replace() #

This is perfect for targeting single or multiple specific values. Just map the string you want to replace to np.nan (don't forget to import numpy first):

import pandas as pd
import numpy as np

# Your original DataFrame
df = pd.DataFrame({
    'type': ['A', 'B', 'C'],
    'size': [1, 1, 1]
})

# Replace 'B' with NaN in the 'type' column
df['type'] = df['type'].replace('B', np.nan)

print(df)

Output:

type  size
0     A     1
1   NaN     1
2     C     1

Method 2: Using mask() #

If you need more conditional flexibility (e.g., replacing based on complex rules), mask() is a great choice. It replaces values where the condition evaluates to True with NaN:

import pandas as pd

# Recreate the original DataFrame
df = pd.DataFrame({
    'type': ['A', 'B', 'C'],
    'size': [1, 1, 1]
})

# Replace 'B' with NaN using a direct condition
df['type'] = df['type'].mask(df['type'] == 'B')

print(df)

This will produce the exact same output as the replace() method.

Bonus: To replace multiple specific strings at once, pass a dictionary to replace():

# Replace both 'B' and 'C' with NaN
df['type'] = df['type'].replace({'B': np.nan, 'C': np.nan})

内容的提问来源于stack exchange，提问作者yangyang