I'm interested in the effect window-to-wall ratio (WWR; ie, the area of windows on a facade divided by the total area of facade) has on the energy consumption of a building. In my energy modelling software, I prepared a base building model representative of the typology my firm typically designs (medium commerical/institutional), and I wrote a script that modifies the WWR on each facade in 20% increments, starting at 20% and progressing to 80%. There are four facades (N, S, E, and W) and each has four WWR alternatives (20%, 40%, 60%, 80%), so I've ended up with 256 models. I've run a simulation for each of these models, and extracted the energy consumption from each into a spreadsheet, which I've attached.

By simply looking at these results I can deduce some clear trends (north-facing glazing increases energy consumption, south-facing glazing does not, etc), but I'd like to be more analytical in my analysis. Unfortunately, the little stats knowledge I have is pretty rusty.

My question is this: is this a valid data set on which multiple linear regression can be applied? Something bothers me about the fact that none of this data is really "random," in that I picked all of the predictor variables, and the response variable is the result of a mathematical calculation (albeit a very complex one).

Any help would be much appreciated.

Thanks,

Josh