Laying the Foundation: Assessing Data Quality in Municipal Building Benchmarking Datasets

摘要

Datasets collected under building benchmarking policies have tremendousrnpotential for changing the way cities and utilities approach energy efficiency.rnComprehensive benchmarking data can be used to understand key trends, target utilityrnprograms, and track progress toward policy goals. However, these uses rely on datarnquality. Benchmarking datasets consist largely of energy data and facility characteristicsrnreported by building owners with little to no training, professional experience, or stake inrnenergy benchmarking. This paper asserts that a process-oriented approach is an importantrnfirst step to understanding data quality, based on analysis of the City of Seattle'srnbenchmarking data.rnA process-oriented approach uses existing datasets and knowledge of data flowsrnto promote deeper understanding of the data and its limitations. The first step in analysisrnof the City of Seattle’s dataset was identifying points where error could be introduced.rnThe research team drew a statistical sample to understand building owner data inputrnpractices. The second step was comparing the Seattle benchmarking dataset withrnmunicipal and third party building datasets to test for inconsistencies. The third step wasrntesting for trends in factors that could explain variability in the data, such as occupancyrnlevels and mixed-use buildings.rnBy testing for sources of error and variability, the authors were able to verify thernoverall quality and utility of the Seattle dataset. The assessment also produced actionablernrecommendations for improvements in benchmarking programs, such as outreachrnstrategies and instructions to building owners.