使用Python3构建推特分类器遇DataFrame属性错误求助

Hey there! Let's work through these errors you're hitting—they're actually pretty common when working with CSV datasets in pandas, so we'll get this sorted quickly.

What's Causing the Errors? #

The AttributeError: 'DataFrame' object has no attribute 'id' and 'friends_count' messages boil down to one simple issue: your DataFrame doesn't have columns with those exact names. This usually happens because:

• The column names in your CSV file are different (e.g., uppercase like Id, spaced like friends count, or camelCase like friendsCount)
• There might be hidden extra spaces or special characters in the column names you didn't notice

First Step: Check Your Actual Column Names #

Since you want to avoid manually digging through the CSV, run this code right after loading your dataset to print all column names automatically:

import pandas as pd

# Load your CSV (replace with your actual file path)
df = pd.read_csv("your_twitter_dataset.csv")
train_df = df.copy()

# Print all column names as a readable list
print("All columns in your dataset:", train_df.columns.tolist())

This will show you the exact column names pandas recognizes. Compare these to the names you're using in your code (id, friends_count)—you'll almost certainly spot a mismatch here.

Fix Your Code to Match Real Column Names #

Once you have the correct column names, use square bracket notation instead of dot notation (this is way more reliable for columns with spaces, special characters, or names that clash with pandas methods). For example:

• If your actual column is named user_id instead of id:

  train_df['user_id'] = train_df['user_id'].apply(lambda x: int(x))
  

• If it's friends count (with a space):

  train_df['friends count'] = train_df['friends count'].apply(lambda x: 0 if x == 'None' else int(x))
  

Bonus: A Cleaner, Faster Way to Handle Numeric Conversions #

Instead of using apply(lambda x: ...), pandas has a built-in pd.to_numeric function that's more efficient and cleaner for this task. It automatically converts invalid values like 'None' to NaN, which you can then fill with 0:

# Convert followers_count to numeric, replace invalid values with 0
train_df['followers_count'] = pd.to_numeric(train_df['followers_count'], errors='coerce').fillna(0).astype(int)

# Do the same for friends_count (no need to run this twice like in your original code!)
train_df['friends_count'] = pd.to_numeric(train_df['friends_count'], errors='coerce').fillna(0).astype(int)

This replaces your two apply lines per column with one concise line, and it's much faster for large datasets.

Quick Note on Redundant Code #

I noticed you're processing friends_count twice in your original code—you can remove one of those lines to avoid unnecessary work.

内容的提问来源于stack exchange，提问作者user12092724