Conclusions: It was concluded that Ridge Regression performed slightly better than other models. It was known that least square method is the model that has the least bias, and highest variance. However, ridge and lasso regressions are using regularization coeffients, which force some of the parameters shrink to zero, increasing bias in the model, as opposed to reduced variance. Since I was aware that regularization eliminate collinearity in the dataset, I was expecting the ridge and lasso prediction results to be better than the least square. However, I was expecting SVR with rbf model to perform better than all of the models, since it is a nonlinear model. Previous studies did not perform Ridge and Lasso regressions in the NYC building data. Therefore, I cannot validate my results with other studies.
Future work: Future work includes expanding this analysis such that other prediction methods will be included, such as decision trees, neural networks etc., The other aspect of the future improvement might be performing a temporal data analysis between 2011 and 2014. In this study I did not perform a temporal analysis, because of the scarcity of the data.
References:
Amasyali, K., & El-Gohary, N. M. (2018). A review of data-driven building energy consumption prediction studies.
Renewable and Sustainable Energy Reviews,
81, 1192-1205. (
Link)
Department of Energy, U.S. (2011). Buildings Energy Databook. Energy Efficiency & Renewable Energy Department. (
Link)
Elith, J., Leathwick, J. R., & Hastie, T. (2008). A working guide to boosted regression trees.
Journal of Animal Ecology,
77(4), 802-813. (
Link)
Ghiaus, C. (2006). Experimental estimation of building energy performance by robust regression.
Energy and buildings,
38(6), 582-587.(
Link)
Kontokosta, C. E. (2015). A market-specific methodology for a commercial building energy performance index.
The Journal of Real Estate Finance and Economics,
51(2), 288-316. (
Link)
Kontokosta, C. E., & Tull, C. (2017). A data-driven predictive model of city-scale energy use in buildings.
Applied Energy,
197, 303-317. (
Link)
Korolija, I., Zhang, Y., Marjanovic-Halburd, L., & Hanby, V. I. (2013). Regression models for predicting UK office building energy consumption from heating and cooling demands.
Energy and Buildings,
59, 214-227. (
Link)
Kontokosta, C. E., & Tull, C. (2017). A data-driven predictive model of city-scale energy use in buildings.
Applied Energy,
197, 303-317. (
Link)
Jain, R. K., Damoulas, T., & Kontokosta, C. E. (2014). Towards Data-Driven Energy Consumption Forecasting of Multi-Family Residential Buildings: Feature Selection via The Lasso. In
Computing in Civil and Building Engineering (2014) (pp. 1675-1682). (
LinkMiller, C., Nagy, Z., & Schlueter, A. (2015). Automated daily pattern filtering of measured building performance data.
Automation in Construction,
49, 1-17 (
Link).