如何在pandas Series中实现‘非’过滤（区别于DataFrame）

Great question! The built-in filter() method for pandas Series only handles positive matching (with like, regex, or in), but there are simple, flexible ways to get the reverse filtering you need. Let's break this down by two common scenarios:

Filtering Out Values #

1. Exclude values matching a string pattern #

Use ~ (the NOT operator) alongside str.contains() to flip the boolean results of a pattern match. This works perfectly for string-based Series:

import pandas as pd

# Sample Series
fruit_series = pd.Series(["apple", "banana", "cherry", "date", "blueberry"])

# Exclude any value containing "berry"
filtered_fruits = fruit_series[~fruit_series.str.contains("berry")]
# Result: 0    apple, 1    banana, 2    cherry, 3    date

Add case=False to ignore capitalization, or na=False to treat NaN values as non-matching (so they get excluded or included based on your logic).

2. Exclude values from a specific list #

If you have a set of exact values to exclude, pair ~ with isin():

exclude_list = ["apple", "date"]
filtered_fruits = fruit_series[~fruit_series.isin(exclude_list)]
# Result: 1    banana, 2    cherry, 4    blueberry

Filtering Out Indexes #

If you need to exclude rows based on their index (instead of values), you have two solid options:

1. Use filter() to get matching indexes, then drop them #

First use the positive filter() to grab the indexes you want to exclude, then remove them with drop():

# Series with labeled indexes
number_series = pd.Series([1, 2, 3, 4, 5], 
                          index=["fruit_apple", "fruit_banana", "veg_carrot", "veg_broccoli", "fruit_cherry"])

# Exclude indexes containing "fruit"
exclude_indexes = number_series.filter(like="fruit").index
filtered_numbers = number_series.drop(exclude_indexes)
# Result: veg_carrot      3, veg_broccoli    4

2. Boolean indexing on the index #

Similar to value filtering, use ~ with index.str.contains() for pattern-based index exclusion:

filtered_numbers = number_series[~number_series.index.str.contains("fruit")]

This also supports regex patterns—just add regex=True and pass your regex string, like ~number_series.index.str.contains(r'^veg_c', regex=True) to exclude indexes starting with "veg_c".

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