如何在Pandas表格中根据指定Series拆分Tag列并展开行？

Got it, let's work through this problem step by step. We need to split the Tag column using a given vocabulary list, keep any non-vocabulary character chunks, and then expand each chunk into its own row with the original ID. Here's how to make it happen:

Step 1: Set up the sample data #

First, let's recreate your input data to make this solution reproducible:

import pandas as pd
import re

# Original DataFrame
df = pd.DataFrame({
    'ID': [1, 2, 3, 4],
    'Tag': ['up down', 'right-left', 'summerfivehigh', 'up a right b left']
})

# Predefined vocabulary Series
vocab = pd.Series(['up', 'down', 'right', 'left', 'summer'])

Step 2: Define the splitting logic #

The core idea is to mark all vocabulary terms in the Tag string with a temporary delimiter, split the string on that delimiter, clean up unwanted whitespace/separators, and retain valid chunks.

We'll use regex to target vocabulary terms, wrap them in a delimiter, then process the split results:

# Create a regex pattern that matches any term in the vocabulary (escaped to handle special characters)
vocab_pattern = re.compile('|'.join(map(re.escape, vocab)))

def split_tag_string(tag):
    # Wrap vocabulary terms with a temporary '|' delimiter to isolate them
    tagged_str = vocab_pattern.sub(r'|\g<0>|', tag)
    # Split on the delimiter, strip extra spaces/hyphens, and filter out empty strings
    split_parts = [part.strip('- ') for part in tagged_str.split('|')]
    # Remove any leftover empty entries from splitting
    cleaned_parts = [part for part in split_parts if part]
    return cleaned_parts

Step 3: Apply the split and expand rows #

Use assign() to apply our splitting function to the Tag column, then explode() to turn each list of chunks into individual rows:

# Apply the split function and expand the results into rows
result_df = df.assign(Tag=df['Tag'].apply(split_tag_string)).explode('Tag').reset_index(drop=True)

print(result_df)

Step 4: Check the output #

Running this code will produce exactly the desired result:

ID       Tag
0   1        up
1   1      down
2   2     right
3   2      left
4   3    summer
5   3  fivehigh
6   4        up
7   4         a
8   4     right
9   4         b
10  4      left

This approach handles all your edge cases: space-separated tags, hyphen-separated tags, concatenated terms like summerfivehigh, and non-vocabulary terms like a and b.

内容的提问来源于stack exchange，提问作者NBC