From PHPMath.com this morning, there's an interesting new posting (lengthy) that talks about implementing a statistical testing library for pulling the chunks of gold out of the raw data from your site.

In this short essay, On Implementing a Statistical Tests Library using PHP, the author offers up his thoughts about the creation of such a library, as well as an overview for the structure of the class. I believe that a folder should be created to hold said code. That folder should be called "Statistics". Underneath that folder there should be at least 3 sub folders: Descriptive, NonParametric, and Parametric. As an example, consider an independent variables T-test.

I would suggest that it might have a class name of Statistics_Parametric_TTest_Independent. This might be shorted to Parametric_TTest_Independent or just TTest_Independent. In the interests of usability, the shorter the class name the better, so ultimately we might want to go with TTest_Independent.

Using some sample algorithmic code from a few other soures, combining them into a larger whole, and creating this class to mine the good bits from your data.

My personal favorite quote: It is a puzzle to me why PHP does not have better math capabilities. I suspect it is partly because there are a lot of Barbies among web developers ("Math is hard. Let's go shopping". - Barbie).