Publication Type

Journal Article

Version

publishedVersion

Publication Date

5-2022

Abstract

Research is needed to explore the limitations and potential for improvement of machine learning for building energy prediction. With this aim, the ASHRAE Great Energy Predictor III (GEPIII) Kaggle competition was launched in 2019. This effort was the largest building energy meter machine learning competition of its kind, with 4370 participants who submitted 39,403 predictions. The test dataset included two years of hourly whole building readings from 2380 meters in 1448 buildings at 16 locations. This paper analyzes the various sources and types of residual model error from an aggregation of the competition’s top 50 solutions. This analysis reveals the limitations for machine learning using the standard model inputs of historical meter, weather, and basic building metadata. The errors are classified according to timeframe, behavior, magnitude, and incidence in single buildings or across a campus. The results show machine learning models have errors within a range of acceptability (RMSLEscaled =  0.3) occur in 4.8% of the test data and are unlikely to be accurately predicted.

Keywords

Building energy prediction, Energy model, Error analysis, Machine learning limitations, Kaggle competition, Artificialintelligence

Discipline

Energy Policy | Engineering

Research Areas

Integrative Research Areas

Publication

Science and Technology for the Built Environment

Volume

28

Issue

5

First Page

610

Last Page

627

ISSN

2374-4731

Identifier

10.1080/23744731.2022.2067466

Publisher

Taylor and Francis Group

Additional URL

https://doi.org/10.1080/23744731.2022.2067466

Download

Share

COinS