如何以Pythonic方式从DataFrame中删除列值出现次数符合条件的行？

Got it, let's work through these two filtering tasks with Pandas. First, here's the original DataFrame we're starting with:

import pandas as pd
data = [['bob', 1], ['james', 4], ['joe', 4], ['joe', 1], ['bob', 3], ['wendy', 5], ['joe', 7]]
df = pd.DataFrame(data, columns=['name', 'score'])
print(df)

Which outputs:

name  score
0   bob      1
1 james      4
2   joe      4
3   joe      1
4   bob      3
5 wendy      5
6   joe      7

Task 1: Remove Rows Where Name Only Appears Once #

The most Pythonic way to handle this is using either value_counts() with isin(), or a more streamlined approach with groupby() + transform(). Here's both options:

Option 1: Using value_counts() #

# Get all names that show up more than once
valid_names = df['name'].value_counts()[df['name'].value_counts() > 1].index
# Filter the DataFrame to keep only those names
filtered_df = df[df['name'].isin(valid_names)]
print(filtered_df)

Option 2: Concise transform() Version #

This avoids creating an intermediate variable and keeps everything in one line:

filtered_df = df[df.groupby('name')['name'].transform('count') > 1]

Either way, you'll get your desired result:

name  score
0   bob      1
2   joe      4
3   joe      1
4   bob      3
6   joe      7

How this works: The transform('count') method adds a temporary column where each row has the total number of times its name appears in the DataFrame. We then filter out any rows where that count is <=1.

Task 2: Remove Rows Where Name Appears 1 or 2 Times #

This is just a small tweak to the previous logic—we now want to keep only names that appear 3 or more times. Here's how to do it:

Option 1: Using value_counts() #

valid_names = df['name'].value_counts()[df['name'].value_counts() >= 3].index
filtered_df = df[df['name'].isin(valid_names)]

Option 2: Concise transform() Version #

filtered_df = df[df.groupby('name')['name'].transform('count') >= 3]

Running this will leave you with only the rows for 'joe', which is the only name that appears 3+ times:

name  score
2   joe      4
3   joe      1
6   joe      7

How this works: We just adjust the threshold from >1 to >=3, so any name with 1 or 2 occurrences gets filtered out.

内容的提问来源于stack exchange，提问作者lordy