In our first Medium post, we detailed a barrier faced by lighting manufacturers and distributors when computing custom rebate estimates for large projects: how do you get good estimates of energy savings? Each project site is different — you might be upgrading from fluorescent tubes, which are already reasonably efficient, and reducing energy usage by about 50%. Or you might be upgrading from inefficient incandescent bulbs and HID fixtures, for a savings of up to 85%.
This week, we upgraded the Rebate Bus API and public QPL tool to better reflect the latest data from our partners. Last time we created an equivalence estimation model, we did so with all categories treated the same. This leads to some errors — typically, an LED T8 is being used to replace a fluorescent T8 while an LED High Bay replaces HID. The savings estimates aught to reflect these common cases. Our previous model didn’t have enough data to learn from to really handle this nuance.
Equivalence estimates from our API are now drawn from a database of thousands of products. We’ve reviewed project data to determine what exactly is the most commonly replaced existing product for all these items, and used LibreOffice Calc’s regression features to produce a prediction model for each product type.
The best part about this is, more users for us means more accurate estimates for you!
Cofounder and CTO, Rebate Bus