Aggregating and Filtering at One Time in SQL

A comprehensive guide to performing data aggregation and filtering simultaneously in SQL for optimized query performance.
SQL
Published

April 6, 2021

When it comes to reporting some statistics by group in SQL, we will think of using the aggregate functions and, if needed, the WHERE clause for filtering. However, the WHERE clause doesn’t allow you to filter on aggregate columns. To complete the query at one time without subqueries, we need the help of the HAVING clause.

Aggregate Functions

There are four popular aggregate functions in SQL:

  • COUNT counts the number of non-NULL rows in a column

    • COUNT(DISTINCT price) counts the number of the distinct non-NULL values in price.

  • SUM adds together all the values in a numeric column

    • NULL is treated as 0.

  • MIN and MAX return the lowest and highest values in a numeric column, respectively

  • AVG calculates the average of a group of selected values

    • It can ONLY be used on numerical columns.

    • NULL is ignored in the calculation.

      SELECT AVG(price)
  FROM tutorial.storesales
 WHERE price IS NOT NULL
---- Equivalent Query ----
SELECT AVG(price)
  FROM tutorial.storesales

Aggregate by Group

For example, you might want to count the number of entries for each year and month.

SELECT year,
       month,
       COUNT(*) AS count
 FROM tutorial.storesales
GROUP BY year, month
ORDER BY month, year

Filtering

What if we are only interested in the months when the highest price is over 200.

SELECT year,
       month,
       MAX(price) AS month_high
 FROM tutorial.storesales
GROUP BY year, month
HAVING MAX(price) > 200
ORDER BY year, month
---- Equivalent Query ----
SELECT *
FROM (
      SELECT year,
             month,
             MAX(price) AS month_high
        FROM tutorial.storesales
       GROUP BY year, month
      ) sub
WHERE month_high > 200
ORDER BY year, month
  • Using the aggregate function and the HAVING clause at one time to filter on an aggregated column.
  • The WHERE clause does not work in this case.
Back to top