Supporting interactive visual analytics of energy behavior in buildings through affine visualizations

Publication: Research - peer-reviewArticle in proceedings


Domain experts dealing with big data are typically not familiar with advanced data mining tools. This especially holds true for domain experts within energy management. In this paper, we introduce a visual analytics approach that empowers such users to visually analyze energy behavior based on consumption meters, sensors and user reported survey data. The approach is aimed at visual analysis of resource consumption data and occupant survey data (e.g. from questionnaires) from apartment buildings. We discuss the principles and architecture of the affine visualization tool, AffinityViz, that interactively maps data from real world buildings. It is an overview +detail inter-active visual analytics tool supporting both rapid ad hoc explorations and structured evaluation of hypotheses about patterns and anomalies in resource consumption data mixed with occupant survey data. We have evaluated the approach with five domain experts within energy management, and further with 10 data analytics experts and found that it was easily attainable and that it supported visual analysis of mixed consumption and survey data. Finally, we discuss future perspectives of affine visual analytics for mixed, explorative visual analysis of resource consumption and occupant survey data.
Original languageEnglish
Title of host publicationProceedings of the 28th Australian Conference on Computer-Human Interaction (OzCHI '16 )
Number of pages10
Place of PublicationNew York, NY, USA
PublisherAssociation for Computing Machinery
Publication year29 Nov 2016
Pages 238-247
ISBN (electronic) 978-1-4503-4618-4
StatePublished - 29 Nov 2016
EventOzCHI '16 - the 28th Australian Conference on Computer-Human Interaction - Launceston, Australia


ConferenceOzCHI '16 - the 28th Australian Conference on Computer-Human Interaction
SeriesOzCHI: Australian Conference on Human-Computer Interaction

See relations at Aarhus University Citationformats

ID: 107883462