“People who watched this also watched” algorithm

Question

I am trying to code an algorithm that acts a bit like Amazon\'s \"People who bought this also bought\".

The difference between the two is that mine is just counting the

Answer 1

I was able to achieve your desired result using a simple DB structure, and a pretty simple query:

Table

TABLE `exa`

| sesh_id | prod_id |
---------------------
| 1       | 1       |
| 1       | 2       |
| 1       | 3       |
| 1       | 4       |
| 2       | 2       |
| 2       | 3       |
| 2       | 4       |
| 3       | 3       |
| 3       | 4       |
| 4       | 1       |
| 4       | 2       |
| 4       | 5       |

Query

SELECT c.prod_id, COUNT(*)
FROM `exa` a
JOIN `exa` b ON a.prod_id=b.prod_id
JOIN `exa` c ON b.sesh_id=c.sesh_id
WHERE a.`prod_id`=3 AND c.prod_id!=3
GROUP BY c.prod_id
ORDER BY 2 DESC;

Result

| prod_id | COUNT |
| 4       | 9     |
| 2       | 6     |
| 1       | 3     |

The idea is that every time a session views a product, it gets inserted into the table [in this case exa]

Then, on any particular product view, you can check and see what other products people who viewed this product also viewed, weighted by frequency. So in this particular example, EVERYONE that viewed product #3 viewed product #4, so it comes up first in the sort. Product #5 was only viewed by session #4, and session #4 didn't view product #3, so product #5 doesn't come up in the results.