Identifying key variables and interactions in statistical models of building energy consumption using regularization

Abstract

a b s t r a c tStatistical models can only be as good as the data put into them. Data about energy consumption continues to grow, particularly its non-technical aspects, but these variables are often interpreted differently among disciplines, datasets, and contexts. Selecting key variables and interactions is therefore an important step in achieving more accurate predictions, better interpretation, and identification of key subgroups for further analysis.This paper therefore makes two main contributions to the modeling and analysis of energy consumption of buildings. First, it introduces regularization, also known as penalized regression, for principled selection of variables and interactions. Second, this approach is demonstrated by application to a comprehensive dataset of energy consumption for commercial office and multifamily buildings in New York City. Using cross-validation, this paper finds that a newly-developed method, hierarchical grouplasso regularization, significantly outperforms ridge, lasso, elastic net and ordinary least squares approaches in terms of prediction accuracy; develops a parsimonious model for large New York City buildings; and identifies several interactions between technical and non-technical parameters for further analysis, policy development and targeting. This method is generalizable to other local contexts, and is likely to be useful for the modeling of other sectors of energy consumption as well.

Exploratory Analysis of Energy Use Across Building Types and Geographic Regions in the United States

“…Two factors contribute to the narrow distribution of values in BPD relative to CBECS. First, the range of values reported in CBECS is likely wide because CBECS captures a diverse cross section of the building stock (Hsu, 2015); the distributions of values in the building stock and in the weighted CBECS dataset are likely narrower than what we observe in the unweighted dataset. Second, BPD may represent only a narrow cross section of the building stock due to self-selection of buildings represented in the dataset.…”

“…A recent increase in building data collection has led to parallel efforts to aggregate (Mathew et al, 2015) and summarize (Kontokosta, 2012a;Hsu, 2014a,b) the available data, and to develop data-driven algorithms for modeling building energy consumption (Kontokosta, 2012b;Hsu, 2014aHsu, , 2015Walter et al, 2014). Mathew et al (2015) and Brown et al (2014) identify the Building Performance Database (BPD) (U.S. Department of Energy, 2015a) as a candidate for supporting data-driven algorithms to inform investments in energy efficiency.…”

Integrative interaction analysis using threshold gradient directed regularization

“…Extensive methodological development and data analysis have been conducted . Promising findings have been made for multiple business and industry problems …”

“…[1][2][3] Promising findings have been made for multiple business and industry problems. 4,5 In high-dimensional interaction analysis, there are two generic paradigms. In marginal analysis, one variable or a small number of variables are analyzed at a time.…”

A Dynamic Model for Indoor Temperature Prediction in Buildings

“…Also the representativeness of the data and the resulting performance of the model in real buildings at normal operating conditions may remain questionable. This is especially the issue when using physical models to generate the data [53].…”