pg_running_stats - mergeable running statistics (Welford/Chan) extension for postgresql

From: Chanukya SDS <chanukyasds(at)gmail(dot)com>
To: pgsql-hackers(at)lists(dot)postgresql(dot)org
Subject: pg_running_stats - mergeable running statistics (Welford/Chan) extension for postgresql
Date: 2025-10-17 05:55:54
Message-ID: CAB4f4B6ga-bjBAmWu2FEjXiR539M6zV2J4EOyOoO5EtPGaTFKQ@mail.gmail.com
Views: Whole Thread | Raw Message | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Hi all,

I’d like to share a new PostgreSQL extension called pg_running_stats.

It implements mergeable, numerically stable running statistics using the
Welford and Chan algorithms.
Unlike the built-in aggregates such as avg(), variance(), and stddev(),
which require scanning the entire dataset, pg_running_stats maintains a
compact internal state that can be updated or merged incrementally.

This makes it well-suited for:

1. streaming or real-time analytics where data arrives continuously,
2. incremental computation over large tables,
3. parallel or distributed queries that need to merge partial aggregates
efficiently.

The extension computes:
mean, variance, standard deviation, skewness, kurtosis, and min/max all in
a single pass.

It’s written entirely in C, depends only on PostgreSQL headers, and builds
cleanly on macOS (Homebrew) and Linux using PGXS.

Source and documentation:
https://github.com/chanukyasds/pg_running_stats

Any feedback, testing, or suggestions for improvement would be very welcome.

Thanks,
Chanukya

Browse pgsql-hackers by date

  From Date Subject
Next Message Chao Li 2025-10-17 06:20:07 Re: Fix an unnecessary cast calling elog in ExecHashJoinImpl
Previous Message Michael Paquier 2025-10-17 05:41:07 Re: Improved TAP tests by replacing sub-optimal uses of ok() with better Test::More functions