About Affordability + Environment Ordering

We have defined this lens as a weighting of certain attributes as follows for each class of products:

Scroll to the bottom of the page for a technical discussion of what these numbers represent.

Theory

Products are ordered in relation to their peers, where peers are all the products meeting your search criteria. If you search for roofing insulation, the peer group in which those insulations will be compared will be smaller than the same product ordered against all of the core and shell insulations. We use relative orderings to provide a descision making tool that can function in the absense of verified environmental and performance data, such as a certified test for the R value of a product.

Each attribute has a calculation associated with it that returns a score from 0 to 1 for a given product relative to it's current peer products. A 1 does not ensure that the product is even particularly good in the area the attribute measures, only that it is the best of its peer group in that area. The lenses we use to present ordered products, such as the lens presented above, are a collection of weights to apply to certain attributes to get an overall score from 0 to 1. It is this score which determines the display order of the product.

Throughout this process, Green2Green works hard to avoid drawing false conclusions. First and foremost in this effort is the handling of uncertainy in the system. As the calcuation of a product's relative ordering score is performed, along with the score we carry along the uncertainy of that score; for instance 0.7 out of 1 plus or minus 0.15 or 15%. This uncertainy can come from the particular source of the data. We also automatically add uncertainty as the data ages; a value entered a year ago has more uncertainy than the one entered this month. At the end of the ordering calculation, scores with more than 20% uncertainty are thrown out. This means that depending on what lens you are ordering products with, a product may simply not show up at all, because the system can not determine its ordering with less than 20% uncertainty. It depends on the value system, because the more heavily a particular value system weights a certain attribute, the more uncertainty in that attribute will cause uncertainty in the entire score.