Question

I have a database that tracks user clicks. If someone clicks on the same thing twice it may produce duplicate entries in the database with the same timestamp.

Is there a way to modify my query to only return results with unique timestamps and not the duplicates.

$stmt1 = $dbConn1->prepare("
    SELECT count(*)
    FROM asset_lookup
    ");

Answer 1

Notes:

• The join to product is 100% equivalent to an INNER join because you use product.product_type_id = 2 in the WHERE clause. Besides that, is there a point to count clicks for products that don’t exist in the product table and only in the asset table (if there are any)?

• The LEFT join to event is 100% equivalent to an INNER join because you use event.activation_date in the WHERE clause.

• The join to tracking_product_lookup seems completely useless.

• The COUNT(*) may give wrong results with outer joins. COUNT(*) counts rows in the intermediate result set and the outer joins produce rows with Nulls in some columns (for example a product with no clicks at all, would still show a count of 1).

• To answer the actual question, if you want to count distinct timestamps, use COUNT(DISTINCT column), not COUNT(*).

The query would be written as:

SELECT product.id, product.name, 
       count(DISTINCT asset_lookup.timestamp)
FROM asset_lookup
  JOIN product ON asset_lookup.product_id = product.id
  JOIN event ON asset_lookup.event_id = event.id
WHERE asset_lookup.event_id = ?
  AND product.product_type_id = 2
  AND asset_lookup.timestamp > event.activation_date
GROUP by asset_lookup.product_id ;