如何在Pandas中计算GroupBy分组非空值计数的中位数？

Here's a straightforward way to solve this problem in Pandas, broken down into simple steps:

Step 1: Count Non-Empty Values per Group #

First, group your DataFrame by the g column. For each group, count how many non-empty entries exist in the val column—we can do this by checking which values aren't empty strings (x != '') and summing those boolean results (since True maps to 1 and False maps to 0).

Step 2: Compute the Median of These Counts #

Once you have the non-empty count for each group, call the .median() method on the resulting Series to get the median value across all groups.

Full Code Example #

Let’s use your sample DataFrame to demonstrate:

import pandas as pd

# Create your sample DataFrame
df = pd.DataFrame({
    'g': [1, 1, 2, 2, 2, 3],
    'val': ['a', '', 'b', '', 'c', '']
})

# Step 1: Calculate non-empty value counts for each group
non_empty_counts = df.groupby('g')['val'].apply(lambda x: (x != '').sum())

# Step 2: Find the median of these counts
median_result = non_empty_counts.median()

# Print results to verify
print("Non-empty counts per group:")
print(non_empty_counts)
# Output:
# g
# 1    1
# 2    2
# 3    0
# Name: val, dtype: int64

print("\nMedian of non-empty counts:", median_result)
# Output: Median of non-empty counts: 1.0

Alternative Concise Syntax #

You can also use .agg() instead of .apply() for a more compact version:

median_result = df.groupby('g')['val'].agg(lambda x: (x != '').sum()).median()

Key Notes #

• This approach automatically includes groups with zero non-empty values (like group 3 in your example), so your median calculation accounts for all groups in the original DataFrame.
• If your "empty" values are NaN instead of empty strings, replace x != '' with x.notna().sum() to count non-null values instead.

内容的提问来源于stack exchange，提问作者mkmostafa