Aggregation

A SELECT over customers returns one row per customer. A SELECT plus GROUP BY country returns one row per country, with the per-country aggregates (count, sum, average, min, max) computed across that country's rows. Aggregation is the operation that turns "many rows" into "one row per group" — the language for summarising data instead of listing it.

This module covers the three pieces: the mechanics of GROUP BY and HAVING, the catalogue of aggregate functions and their non-obvious corners (especially around NULL), and the multi-dimensional grouping operators (ROLLUP, CUBE, GROUPING SETS) that compute subtotals at multiple levels in one query.

Place in the curriculum

• Prerequisites: Foundations and Joins. GROUP BY is a step in the logical execution order, and most aggregation queries operate on joined tables.
• Followed by: Window Functions. Window functions generalise aggregation — you can keep one row per original row while computing aggregates across a window of related rows.

Chapters

1. GROUP BY and HAVING — the mechanics: how rows collapse into groups, what columns are legal in SELECT after a GROUP BY, and the difference between WHERE (filters rows before grouping) and HAVING (filters groups after).
2. Aggregate Functions — COUNT, SUM, AVG, MIN, MAX, STRING_AGG/GROUP_CONCAT, plus the modifiers (DISTINCT, FILTER) that change their meaning. Plus the NULL behaviour that surprises everyone the first time.
3. Grouping Sets, ROLLUP, CUBE — multi-dimensional aggregation in a single query: subtotals by country, subtotals by month, subtotals by both. The right tool for spreadsheet-style "what does the data look like at every level."