Analyzing Related Items

[kjs50]

Hello,

I am searching for any type of software that has some sort of artifical intelliegence to recognize patterns. I want to show my customers what items other people ordered with a specific item. I would like to go through my past orders, mark a few as "good" combinations and "bad" combinations (where the customer bought random stuff not related to the main item). And then afterwards automatically display other items of interest to the customer based on recent orders.

Sounds like Bayesian filtering, but wondering if anyone else has any ideas or software solutions readymade.

Thanks.

[jwolthuis]

I implemented this on my web site, but didn't do anything fancy. I query previous orders, and sort descending by quantity.

After several months of solid data, "good combinations" bubble to the top of the list.

[saicam]

I would suggest voogo it does exactly what you want. With help from vogoo, you can give people recommendations based on their tastes, and implement "who liked this also liked" features.

I didn't try it personally but have read positive things about it. There are two versions of the software a PHP and C++ one. There are a couple of tutorials and a demo on the site, so you can check if it suits You needs.

You may also do a search for collaborative filtering, on the wikipedia page (http://en.wikipedia.org/wiki/Collaborative_filtering) for collaborative filtering You can find some interesting links to this subject.

I wanted to implement some of the collaborative search algorithms in the ecommerce site I was working on recently, but the "dead line" came to early (as always ;) and I haven't managed to implement that. Hope you have more luck :)

[edited by: lorax at 3:33 pm (utc) on Nov. 24, 2005]
[edit reason] Removed URL [/edit]