Python中如何将数据列从Object类型统一转换为int64类型？

Got it, let's work through converting your mixed Object type Result column to int64 step by step. Your data has a mix of values—some with units (like "98.8 PG/ML H"), pure decimal strings, and non-numeric entries like "CREDIT"—so we need to clean the data first before changing the type.

Step 1: Extract numeric values from messy strings #

First, we'll use regular expressions to pull out the numeric part from each entry. This handles cases where units are attached to numbers. We'll write a helper function for this:

import pandas as pd
import re

def extract_num(val):
    # Match integers or decimals (supports positive/negative values too)
    num_match = re.search(r'[-+]?\d*\.\d+|\d+', str(val))
    if num_match:
        # Convert the matched string to float first (since we have decimals)
        return float(num_match.group())
    else:
        # Return NaN for entries with no numeric data (like "CREDIT")
        return pd.NA

# Apply the function to your Result column
data['Result'] = data['Result'].apply(extract_num)

Step 2: Handle non-numeric entries #

After extracting, you'll have NaN values where there was no number (like the "CREDIT" entry). You need to decide how to handle these based on your use case:

• Replace with 0 (common if missing/non-numeric means 0):

  data['Result'] = data['Result'].fillna(0)
  

• Or keep NaN and convert to Int64 (capital I, which supports nullable integers) instead of int64 if you want to preserve missing values:

  data['Result'] = data['Result'].astype('Int64')
  

Step 3: Convert to int64 #

If you've filled NaN with a numeric value (like 0), you can now convert to int64. Note that we first converted to float to handle decimals—direct conversion from string to int would fail for decimal values:

# Optional: Round decimals if you want the nearest integer instead of truncating
data['Result'] = data['Result'].round()

# Convert to int64
data['Result'] = data['Result'].astype('int64')

Testing with your sample data #

Let's see what happens to your example entries:

• "98.8 PG/ML H" → 98.8 → rounded to 99 (or truncated to 98 if you skip rounding) → int64
• "8.20000" → 8.2 → 8 or 9 → int64
• "26.8 PG/ML H" →26.8 →27 or 26 → int64
• "40.8 PG/ML H" →40.8 →41 or 40 → int64
• "CREDIT" → NaN → 0 (if you filled) → int64
• "15.30000" →15.3 →15 or 16 → int64

That should get your column into a clean int64 format!

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