Energy efficiency program evaluators use data from a variety of sources, which range from utility billing databases to surveys to logs from smart thermostats. We estimate reductions in energy usage attributable to various energy savings programs. Our evaluations are used to certify that utilities are meeting state or federally mandated efficiency standards; to reimburse utilities for funds spent on what amounts to reducing demand for power; or to assess whether programs are worth continuing or expanding. It is therefore critical that our estimates be reliable and reproducible. nnThe raw data we receive is usually very messy, and the types of problems we find in the data are sufficiently niche that off-the-shelf data cleaning software is of limited help. nnOur poster demonstrates a suite of custom R packages and Shiny apps that we've developed to streamline the process of prepping energy use data for analysis. Our current tools include:n• A package, noaaisd, which pulls appropriate hourly weather data from the NOAA website and appends it to geocoded customer usage datan• A Shiny app that allows users to quickly explore individual smart thermostat logs and save information about the quality of each log’s data nnFeatures under development include:n• A package to automate the cleaning of utility billing data (this requires, among other things, the ability to detect and appropriately correct gaps or overlaps in billing periods for individual customers, and the ability to flag abnormal billing periods or energy consumption)n• A Shiny app to help build baseline energy usage models
