如何删除以特定通配符字符串结尾的DataFrame列？

First, let's fix the issues in your original code:

• You tried to call .str.contains() on individual string column names (c), which isn't valid—.str is a pandas accessor for Series/Index objects, not single strings.
• Your regex pattern [a+z] was incorrect; it should be [a-z] to match any lowercase letter (the + was a typo here).

Here are two working approaches to achieve your goal:

Approach 1: Using Pandas' Built-in String Methods #

This is the most concise way, leveraging pandas' Index.str methods to filter columns:

# Filter columns that match the pattern: publication_date_ followed by a lowercase letter at the end
columns_to_drop = df_merge.columns[df_merge.columns.str.contains(r'publication_date_[a-z]$')]

# Drop the identified columns
df_merge = df_merge.drop(columns_to_drop, axis=1)

Regex Breakdown: #

• publication_date_: Exact match for the prefix
• [a-z]: Matches any single lowercase letter
• $: Ensures the string ends right after the letter (so no extra characters are present)

Approach 2: Using the re Module for Explicit Matching #

If you prefer using Python's standard regex library:

import re

# Iterate through columns and collect those matching the pattern
columns_to_drop = [col for col in df_merge.columns if re.fullmatch(r'publication_date_[a-z]', col)]

# Drop the columns
df_merge = df_merge.drop(columns_to_drop, axis=1)

Why This Works: #

• re.fullmatch() checks if the entire column name matches the pattern, which is equivalent to adding $ in the previous regex.
• This will target exactly columns like publication_date_x and publication_date_y, leaving publication_date (without the suffix) intact.

After running either of these, your DataFrame will have the desired columns: contract_number, award_date, award_date_x, award_date_y, publication_date.

内容的提问来源于stack exchange，提问作者Mitchell