Well, you’ve conjured up quite a project for yourself. To make a script specific to your needs you’ll probably need to know a language like PHP and have access to a database like MYSQL. You’ll also need a ton of data to fill that database with – you can’t test what you don’t have. All in all, getting that data is pretty hard. You could of course make up fake data and go about it that way, but all in all, I think if you don’t know any programming languages then you’re kinda screwed.

Oh, I see in your profile that you already have the PHP stuff down :p

I am very fluent in PHP and MySQL. I have a bunch of data to play with, but I’m not sure what to do with it. How do I determine what to recommend? What methods are there to go about this.

My head is about to explode from this, :)

Any suggestions would be much appreciated!

Does the data have tags or descriptions?

I have to head out, but I’ll leave you with this.
If you can somehow give the data tags or descriptions from which you can harvest tags then you’re up a creek.
Another alternative is if the data has a list of users who bought what then you could compare which users bough the two same items and from that you can show a recommendation for each project.

It has descriptions but I can easily make tags for them that shouldn’t be too hard. I’m really just looking for ideas..

Ah, what kind of ideas? Layout, etc. or…?

on a method of determining what to auggest

Determine it based on how close the two tags are related. They’d have to be the same, or how many tags the two products share.

The algorithm for proper recommendation is quite scientific and you’ll need to read a few papers on it (link). There are currently several popular approaches. One is based on buying history analysis and the other is based on the browsing history analysis. They are similar and go like this:

The history shows that one customer looked at products A, B, and C and purchased B. Another customer that purchased B also looked at C. Another customer that purchased B also looked at A. And so on. You count the hits for each product combo within a browsing or buying session and that gives you relevance metric to determine that products are related or not.