Energy Efficiency

, Volume 12, Issue 8, pp 2055–2078 | Cite as

Large scale residential energy efficiency prioritization enabled by machine learning

  • Badr Al TarhuniEmail author
  • Adel Naji
  • Philip G. Brodrick
  • Kevin P. Hallinan
  • Robert J. Brecha
  • Zhongmei Yao
Review Article


Cost-effective energy efficiency improvements in residential buildings can realize 30% energy reduction within this energy sector for the USA. Unfortunately, audits on residences to identify potential savings, generally based on detailed energy models, tend to over-estimate anticipated savings. This leads to wariness on the part of potential investors. We address this issue by taking an integrated data- and physics-based approach, using residential building geometrical and energy system characteristics (e.g., envelope, appliances, and HVAC systems), as well as historical energy consumption for each residence and weather data, to predict monthly natural gas energy consumption and savings from the adoption of individual energy saving measures. This approach requires only those geometrical and energy system characteristics associated with the greatest potential for realized savings which are also easy to obtain. We construct a dataset from a collection of houses with a wide range of energy effectiveness, and train a single statistical model that accurately predicts natural gas energy consumption in any of the individual residences. The model is then used to estimate the savings from most impactful energy efficiency investments for individual residences, as well as the collective grouping. The specific case considered here involves hundreds of university-owned student residences in the U.S. Midwest. The resulting machine-learning derived model is used to predict monthly natural gas consumption with a mean squared error of 0.00023 for unitary scaled cross-validation data. We use a nearest neighbor approach to validate savings estimates for virtually improved residences, identifying the surrogate residence most like an improved residence. This validation demonstrates a savings prediction accuracy to within 3.5% for most of the measures considered. The validated models are shown capable of prioritizing investments among the collective of residences considered. Sequential adoption of the most cost-effective energy efficiency measure among a group of residences renders a total energy reduction of 36%. The practical implications of this research are significant. This integrated machine-learning and physics-based approach to estimate savings could be used in any utility district and throughout all of the USA to enable district-wide residential energy reduction based upon sequential adoption of the most cost-effective energy measures.


Machine learning Data-based energy models Residential Energy savings Prioritized reduction 



  1. Ahmad, T., & Chen, H. (2018). Short and medium-term forecasting of cooling and heating load demand in building environment with data-mining based approaches. Energy & Buildings, 166, 460–476.CrossRefGoogle Scholar
  2. Al-Qahtani, F. H., & Crone, S. F. (2013). Multivariate k-Nearest Neighbour Regression for Time Series data -a novel Algorithm for Forecasting UK Electricity Demand. Neural Networks (IJCNN), The 2013 International Joint Conference on. Google Scholar
  3. Aqlan, F., Ahmed, A., Srihari, K., & Khasawneh, M. T. (2014). Integrating artificial neural networks and cluster analysis to assess energy efficiency of buildings. 2014 Industrial and Systems Engineering Research Conference. Canada: . Khasawneh.Google Scholar
  4. Aydinalp, M., Ugursal, I. V., & Fung, F. S. (2002). Modeling of the appliance, lighting, and space-cooling energy consumptions in the residential sector using neural networks. Applied Energy, 71, 87–110.CrossRefGoogle Scholar
  5. Baty, F., Ritz, C., Brutsche, S., Flandrois, J.-P., & Delignette-Muller, M.-L. (2015). A toolbox for nonlinear regression in {R}: The package nlstools. Journal of Statistical Software, 66(5), 1–21.CrossRefGoogle Scholar
  6. Brecha, R. J., Mitchell, A., Hallinan, K. P., & Kissock, J. K. (2011). Prioritizing investment in residential energy efficiency and renewable energy - a case study for the U.S. Midwest. Energy Policy, 39(5), 2982–2992.CrossRefGoogle Scholar
  7. Chonan, Y., Nishida, K., & Matsumoto, T. (1996). Great energy predictor shootout II–A Bayesian nonlinear regression with multiple hyperparameters. 1996 Annual Meeting of the American-Society-of-Heating-Refrigerating-and-Air-Conditioning-Engineers, Inc, 102, pp. 405–411. San Antonio.Google Scholar
  8. Deakins, D. (2010). Energy modeling using feed-forward artificial neural networks. 2010 Undergraduate Research and Scholarship, (pp. 4–12).Google Scholar
  9. Drehobl, A., & Ross, L. (2016). Lifting the High Energy Burden in America’s Largest Cities: How Energy Efficiency Can Improve Low Income and Underserved Communities. ACEEE.Google Scholar
  10. Duncan, J., Ballinger, M., Fritz, B., Becker, J., Tilden, H., Raney, E., ... Mendez, K. a-C. (2013). Pacific Northwest National Laboratory Annual Site Environmental Report for Calendar Year 2012. Washington.Google Scholar
  11. EIA. (2017). Energy Information Administration (EIA) . Accessed 7 June 2018
  12. Ekici, B. B., & Aksoy, U. T. (2009). Prediction of building energy consumption by using artificial neural networks. Advances in Engineering Software, 40(5), 356–362.CrossRefGoogle Scholar
  13. ENERGY STAR. (2017). ENERGY STAR - Energy Efficient Products. Accessed September 2016
  14. EPA. (2016). United States Environmental Protection Agency-Greenhouse Gas Emissions. Accessed 10 June 2018
  15. Fels, M. (1986). PRISM: an introduction. Energy and Buildings, 9(1–2), 5–18.CrossRefGoogle Scholar
  16. Frankel, M., & Turner, C. (2008). Energy performance of LEED NC buildings. NBI.Google Scholar
  17. Granade, H. C., Creyts, J., Derkach, A., Farese, P., Nyquist, S., & Ostrowski, K. (2009). Unlicking energy efficiency in the U.S economy. McKinsey Global Energy and Materials.Google Scholar
  18. Higgins, A., Foliente, G., & McNamara, C. (2011). Modelling intervention options to reduce GHG emissions in housing stock—A diffusion approach. Technological Forecasting and Social Change, 78(4), 621–634.CrossRefGoogle Scholar
  19. Karatasou, S., Santamouris, M., & Geros, V. (2006). Modeling and predicting building’s energy use with artificial neural networks: Methods and results. Energy and Buildings, 38(8), 949–958.CrossRefGoogle Scholar
  20. Kind, P. H. (2015). Pathway to a 21st century electric utility. Boston, MA: CERES.Google Scholar
  21. Kreider, J., Claridge, D., Curtiss, P., Dodier, R., Haberl, J., & Krarti, M. (1995). Building energy use prediction and system identification using recurrent neural networks. Journal of Solar Energy Engineering, 117(3), 161–166.CrossRefGoogle Scholar
  22. Kuusinen, S. (2014). Guiding building owners towards energy conservation measures. University of Colorado Boulder.Google Scholar
  23. Liaw, A., & Wiener, M. (2002). Classification and regression by randomForest. R News, 2(3), 18–22. Retrieved from Scholar
  24. Neto, A. H., & Fiorelli, F. v. (2008). Comparison between detailed model simulation and artificial neural network for forecasting building energy consumption. Energy and Buildings, 40(12), 2169–2176.CrossRefGoogle Scholar
  25. Nizami, S. S., & Z AI-Garni, A. (1995). Forecasting electric energy consumption using neural networks. Energy Policy, 23(12), 1097–1104.CrossRefGoogle Scholar
  26. NOAA. (2016, 5 31). National Center for environmental information. Retrieved from Climate Data Online:
  27. Parmar, A. A. (2015). Deep Learning with H2O (3rd ed.). Mountain View, CA:, Inc..Google Scholar
  28. Pigg, S. (2003). Electricity Use by New Furnaces. Madison: focus on energy.Google Scholar
  29. R Core Team. (2013). R: A language and environment for statistical computing. R Foundation for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing. Retrieved from
  30. Schliep, K., Hechenbichler, K., & A.Lizee. (2016). Weighted k-Nearest Neighbors. Retrieved from L
  31. Stone, M. (1974). Cross-validatory choice and assessment of statistical prediction. Journal of the Royal Statistical Society, 111–147.MathSciNetzbMATHGoogle Scholar
  32. Ternes, M. P., & and Gettings, M. B. (2008). Analyses to Verify and Improve the Accuracy of the Manufactured Home Energy Audit (MHEA). Oak Ridge, Tennessee: Energy and Transportation Science Division.Google Scholar
  33. Tsanas, A., & Xifara, A. (2012). Accurate quantitative estimation of energy performance of residential buildings using statistical machine learning tools. Energy and Buildings, 49, 560–567.CrossRefGoogle Scholar
  34. Villoria-Siegert, R., Brodrick, P., Hallinan, K. P., & Brecha, R. (2014). Cost-availability curves for hierarchical implementation of residential energy-efficiency measures. Energy Efficiency, 8(2), 267–279.CrossRefGoogle Scholar
  35. Wickham, H. (2017). ggplot2: Elegant graphics for data analysis. New York: Springer-Verlag New York Retrieved from Scholar
  36. Yu, W., LI, B., Lei, Y., & Liu, M. (2011). Analysis of a residential building energy consumption demand model. energies, 4(3), 475–487.CrossRefGoogle Scholar

Copyright information

© Springer Nature B.V. 2019

Authors and Affiliations

  • Badr Al Tarhuni
    • 1
    Email author
  • Adel Naji
    • 1
  • Philip G. Brodrick
    • 2
  • Kevin P. Hallinan
    • 3
  • Robert J. Brecha
    • 4
  • Zhongmei Yao
    • 5
  1. 1.Department of Mechanical and Aerospace Engineering/Renewable and Clean Energy, School of EngineeringUniversity of DaytonDaytonUSA
  2. 2.Department of Global EcologyCarnegie Institution for SciencePalo AltoUSA
  3. 3.Department of Mechanical and Aerospace Engineering/Renewable and Clean Energy, Building Energy CenterUniversity of DaytonDaytonUSA
  4. 4.Department of Physics and Renewable and Clean Energy Program, Hanley Sustainability InstituteUniversity of DaytonDaytonUSA
  5. 5.Department of Computer ScienceThe University of DaytonDaytonUSA

Personalised recommendations