In late 2019, ASHRAE hosted the Great Energy Predictor III (GEPIII) machine learning competition on the Kaggle platform. This launch marked the third energy prediction competition from ASHRAE and the first since the mid-1990s. In this updated version, the competitors were provided with over 20 million points of training data from 2,380 energy meters collected for 1,448 buildings from 16 sources. This competition’s overall objective was to find the most accurate modeling solutions for the prediction of over 41 million private and public test data points. The competition had 4,370 participants, split across 3,614 teams from 94 countries who submitted 39,403 predictions. In addition to the top five winning workflows, the competitors publicly shared 415 reproducible online machine learning workflow examples (notebooks), including over 40 additional, full solutions. This paper gives a high-level overview of the competition preparation and dataset, competitors and their discussions, machine learning workflows and models generated, winners and their submissions, discussion of lessons learned, and competition outputs and next steps. The most popular and accurate machine learning workflows used large ensembles of mostly gradient boosting tree models, such as LightGBM. Similar to the first predictor competition, preprocessing of the data sets emerged as a key differentiator.

Cities across the country (20 to date) are rapidly passing laws to mandate the collection and disclosure of energy usage data with the hopes that such data could be utilized to benchmark building energy performance and provide a basis for designing and deploying efficiency measures. However, numerous municipalities are struggling to translate such data into actionable insights on which buildings are the best candidates for energy efficiency interventions. Current studies are limited in their ability to compare energy efficiency of buildings at the city scale and maintain interpretability necessary to result in effective decision making that can facilitate municipal policy and program design. In this paper, we present an integrated data-driven method based on recursive partitioning and stochastic frontier analysis to benchmark energy performance of building stock across an entire city. We implement the proposed method on a dataset of over 10,000 buildings in New York City. The preliminary results begin to quantify the potential for building energy efficiency and give further support to city officials for examining effects of possible interventions. We aim to establish a robust methodology that provides a standard way to benchmark building energy performance at the city scale, which can be easily applied to other cities across the country.

Cities across the country (20 to date) are rapidly passing laws to mandate the collection and disclosure of energy usage data with the hopes that such data could be utilized to benchmark building energy performance and provide a basis for designing and deploying efficiency measures. We present an integrated data-driven method based on recursive partitioning and stochastic frontier analysis to benchmark energy performance of the building stock across an entire city. We implement the proposed method to over 10,000 buildings in New York City. The preliminary results quantify the potential for building energy efficiency and support policymakers for further examining effects of interventions.