Linear regression is a statistical tool that you can use to compare the relationship between two variables. This technique finds its usefulness in project estimating in the form of comparing a cost driver to the project' s cost.
A cost driver is any item, process, or event that has a direct relation to a cost. A server, an added human resource, a software program, a network connection—all of these are cost drivers. These are obvious, but a cost driver might be a fairly esoteric thing. Here ' s an example: For years, there was a rumor floating around in IT circles that a random access memory (RAM) factory somewhere in Asia had burned down, and that this explained why the cost of memory was so high. If you were an expert in the RAM business, perhaps you would 've been able to validate or disprove this rumor, but if you were a RAM buyer, the price was high enough to justify the rumor. This rumor, itself, was a cost driver. There are other cost drivers in IT—some put in place to keep prices high, and still others due to a technological shortcoming of some kind.
Linear regression tries to make sense of cost drivers to dependent costs. Dor example, the price of RAM is high, therefore that has an input to the price of the PCs we build—the price of our PCs are, to some extent, dependent on the price of the RAM we put in them.
Was this article helpful?