Making sense of building data: New analysis methods for understanding indoor climate. (15th January 2018)

• Record Type: Journal Article
• Authors: Geronazzo, Angela; Brager, Gail; Manu, Sanyogita
• Is Part Of: Building and environment. Volume 128(2018)
• Page Start: 260
• Page End: 271
• Publication Date: 2018-01-15
• Subjects: Passive buildings -- Field data -- Decision trees -- Comfort -- Data representation; Buildings -- Environmental engineering -- Periodicals; Building -- Research -- Periodicals; Constructions -- Technique de l'environnement -- Périodiques; Electronic journals; 696
• DOI: 10.1016/j.buildenv.2017.11.030
• Languages: English
• ISSNs: 0360-1323

Abstract

Abstract: This work presents a novel approach for exploratory analysis of indoor environment measurements. Building monitoring is a complex task constrained by technological and environmental limits. Monitoring processes can be long, employ multiple sensors and equipment, and may involve multiple buildings across extended periods. Once the data is collected, it's not always straightforward how to make sense of it. When faced with complex heterogeneous amounts of data, early exploratory analysis is helpful to make sense of the broad patterns, highlight data weaknesses and features, and identify directions for more in-depth analysis. Inaccuracy, incompleteness and inconsistence are common characteristics of monitored field data, which then often needs intensive pre-processing to address data quality issues and improve data analysis results. This paper describes techniques from computer science, such as decision tree induction and others, that are not normally applied to field measurements of indoor environments in buildings, and as an example applies them to a complex set of data from climate-responsive spaces in India. The findings illustrate the power of these techniques for using simplified measures as proxies to extract more sophisticated information about building operation, and for quickly identifying patterns in how different design and operational characteristics affect thermal comfort. Highlights: Field-base dataset exploration and representation. Approximation methods for unavailable data. Application of decision tree induction to environmental building data. Definition of relevant feature attributes with reference to perceived comfort.