{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/learning-short-term-past-as-predictor-of","title":"Learning short-term past as predictor of human behavior in commercial buildings","arxiv_id":"1809.10020","date":"2018-09-17","proceeding":null,"authors":["Romana Markovic","Jérôme Frisch","Christoph van Treeck"],"abstract":"This paper addresses the question of identifying the time-window in\nshort-term past from which the information regarding the future occupant's\nwindow opening actions and resulting window states in buildings can be\npredicted. The addressed sequence duration was in the range between 30 and 240\ntime-steps of indoor climate data, where the applied temporal discretization\nwas one minute. For that purpose, a deep neural network is trained to predict\nthe window states, where the input sequence duration is handled as an\nadditional hyperparameter. Eventually, the relationship between the prediction\naccuracy and the time-lag of the predicted window state in future is analyzed.\nThe results pointed out, that the optimal predictive performance was achieved\nfor the case where 60 time-steps of the indoor climate data were used as input.\nAdditionally, the results showed that very long sequences (120-240 time-steps)\ncould be addressed efficiently, given the right hyperprameters. Hence, the use\nof the memory over previous hours of high-resolution indoor climate data did\nnot improve the predictive performance, when compared to the case where 30/60\nminutes indoor sequences were used. The analysis of the prediction accuracy in\nthe form of F1 score for the different time-lag of future window states dropped\nfrom 0.51 to 0.27, when shifting the prediction target from 10 to 60 minutes in\nfuture.","url_abs":"http://arxiv.org/abs/1809.10020v1","url_pdf":"http://arxiv.org/pdf/1809.10020v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"learning-short-term-past-as-predictor-of","repo_url":"https://github.com/littlejiao/figureextract","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}