{"id":{"repo_id":"exeter","oai_identifier":"oai:figshare.com:article/31970667"},"canonical_url":"https://search.dev.ndltd.org/etd/exeter/oai:figshare.com:article/31970667","repository":{"repo_id":"exeter","name":"University of Exeter","base_url":"https://api.figshare.com/v2/oai"},"display":{"title":"Bayesian Optimisation of Hyperparameters in Regression Models for Smart Energy and Environmental Systems","abstract":"The 2030 Agenda for Sustainable Development Goals (SDGs) aims to curb global warming and air pollution by improving the efficiency of sustainable, low-emission energy production and consumption in smart homes and smart cities. Achieving this requires the development of advanced methods to enhance energy generation and consumption within rapidly growing hybrid energy systems. One effective approach is to improve the forecasting of renewable energy production, energy demand, and air quality in smart environments. Therefore, developing data-driven methods to exploit the increasing volume of data generated by smart homes and smart cities is essential for improving efficiency and reducing environmental impact. This thesis presents three case studies focusing on air-quality prediction, household energy demand modelling, and renewable energy production forecasting. The results demonstrate that hyperparameter optimisation significantly improves predictive performance and computational efficiency. For example, optimisation of artificial neural network models improved prediction accuracy for smart home energy consumption from 60\\% to approximately 85\\%, representing an improvement of around 42\\%. Similarly, in modelling energy consumption, predictive performance increased from 0.52 to 0.73, corresponding to an improvement of approximately 40\\%. Gaussian process regression models also achieved high predictive accuracy while reducing computational cost through efficient hyperparameter selection. In addition, the findings reveal a consistent trade-off between prediction accuracy and computational time, highlighting the importance of balanced optimisation strategies for real-world applications. This thesis makes a novel contribution by developing a user-preference-based framework for Bayesian hyperparameter optimisation across smart energy and environmental applications. The proposed approach enables systematic tuning of model hyperparameters to balance accuracy and computational cost according to application requirements. Overall, this work demonstrates how data-driven optimisation can enhance predictive performance and support the development of efficient and sustainable smart energy and environmental systems.<p></p>","abstract_html":"The 2030 Agenda for Sustainable Development Goals (SDGs) aims to curb global warming and air pollution by improving the efficiency of sustainable, low-emission energy production and consumption in smart homes and smart cities. Achieving this requires the development of advanced methods to enhance energy generation and consumption within rapidly growing hybrid energy systems. One effective approach is to improve the forecasting of renewable energy production, energy demand, and air quality in smart environments. Therefore, developing data-driven methods to exploit the increasing volume of data generated by smart homes and smart cities is essential for improving efficiency and reducing environmental impact. This thesis presents three case studies focusing on air-quality prediction, household energy demand modelling, and renewable energy production forecasting. The results demonstrate that hyperparameter optimisation significantly improves predictive performance and computational efficiency. For example, optimisation of artificial neural network models improved prediction accuracy for smart home energy consumption from 60\\% to approximately 85\\%, representing an improvement of around 42\\%. Similarly, in modelling energy consumption, predictive performance increased from 0.52 to 0.73, corresponding to an improvement of approximately 40\\%. Gaussian process regression models also achieved high predictive accuracy while reducing computational cost through efficient hyperparameter selection. In addition, the findings reveal a consistent trade-off between prediction accuracy and computational time, highlighting the importance of balanced optimisation strategies for real-world applications. This thesis makes a novel contribution by developing a user-preference-based framework for Bayesian hyperparameter optimisation across smart energy and environmental applications. The proposed approach enables systematic tuning of model hyperparameters to balance accuracy and computational cost according to application requirements. Overall, this work demonstrates how data-driven optimisation can enhance predictive performance and support the development of efficient and sustainable smart energy and environmental systems.&lt;p&gt;&lt;/p&gt;","abstract_has_math":false,"creators":["Ahmed Alzimami (21059816)"],"institution":null,"degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":[],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2026,"date_issued":"2026-04-13T00:00:00Z","date_published":"2026-04-13T00:00:00Z","updated_at":"2026-07-27T19:33:32Z","subjects":["Tuned Regression","Smart Home Energy"],"languages":[],"rights":["All rights reserved","Open Access after 2027-10-13"],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier","label":"Identifier","values":["10779/exe.31970667.v1"],"render_values":[{"text":"10779/exe.31970667.v1","href":null,"code":true}]}]},"links":{"outbound_url":null,"outbound_label":null,"outbound_source":null},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:creator","label":"Author","values":["Ahmed Alzimami (21059816)"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2026-04-13T00:00:00Z"]},{"key":"dc:relation","label":"Dc Relation","values":["https://figshare.com/articles/thesis/Bayesian_Optimisation_of_Hyperparameters_in_Regression_Models_for_Smart_Energy_and_Environmental_Systems/31970667"]},{"key":"dc:type","label":"Dc Type","values":["Text","Thesis"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Tuned Regression","Smart Home Energy"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:rights","label":"Dc Rights","values":["All rights reserved","Open Access after 2027-10-13"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["10779/exe.31970667.v1"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["The 2030 Agenda for Sustainable Development Goals (SDGs) aims to curb global warming and air pollution by improving the efficiency of sustainable, low-emission energy production and consumption in smart homes and smart cities. Achieving this requires the development of advanced methods to enhance energy generation and consumption within rapidly growing hybrid energy systems. One effective approach is to improve the forecasting of renewable energy production, energy demand, and air quality in smart environments. Therefore, developing data-driven methods to exploit the increasing volume of data generated by smart homes and smart cities is essential for improving efficiency and reducing environmental impact. This thesis presents three case studies focusing on air-quality prediction, household energy demand modelling, and renewable energy production forecasting. The results demonstrate that hyperparameter optimisation significantly improves predictive performance and computational efficiency. For example, optimisation of artificial neural network models improved prediction accuracy for smart home energy consumption from 60\\% to approximately 85\\%, representing an improvement of around 42\\%. Similarly, in modelling energy consumption, predictive performance increased from 0.52 to 0.73, corresponding to an improvement of approximately 40\\%. Gaussian process regression models also achieved high predictive accuracy while reducing computational cost through efficient hyperparameter selection. In addition, the findings reveal a consistent trade-off between prediction accuracy and computational time, highlighting the importance of balanced optimisation strategies for real-world applications. This thesis makes a novel contribution by developing a user-preference-based framework for Bayesian hyperparameter optimisation across smart energy and environmental applications. The proposed approach enables systematic tuning of model hyperparameters to balance accuracy and computational cost according to application requirements. Overall, this work demonstrates how data-driven optimisation can enhance predictive performance and support the development of efficient and sustainable smart energy and environmental systems.<p></p>"]},{"key":"dc:title","label":"Title","values":["Bayesian Optimisation of Hyperparameters in Regression Models for Smart Energy and Environmental Systems"]}]}],"canonical_facts":{"dc:creator":["Ahmed Alzimami (21059816)"],"dc:date":["2026-04-13T00:00:00Z"],"dc:description":["The 2030 Agenda for Sustainable Development Goals (SDGs) aims to curb global warming and air pollution by improving the efficiency of sustainable, low-emission energy production and consumption in smart homes and smart cities. Achieving this requires the development of advanced methods to enhance energy generation and consumption within rapidly growing hybrid energy systems. One effective approach is to improve the forecasting of renewable energy production, energy demand, and air quality in smart environments. Therefore, developing data-driven methods to exploit the increasing volume of data generated by smart homes and smart cities is essential for improving efficiency and reducing environmental impact. This thesis presents three case studies focusing on air-quality prediction, household energy demand modelling, and renewable energy production forecasting. The results demonstrate that hyperparameter optimisation significantly improves predictive performance and computational efficiency. For example, optimisation of artificial neural network models improved prediction accuracy for smart home energy consumption from 60\\% to approximately 85\\%, representing an improvement of around 42\\%. Similarly, in modelling energy consumption, predictive performance increased from 0.52 to 0.73, corresponding to an improvement of approximately 40\\%. Gaussian process regression models also achieved high predictive accuracy while reducing computational cost through efficient hyperparameter selection. In addition, the findings reveal a consistent trade-off between prediction accuracy and computational time, highlighting the importance of balanced optimisation strategies for real-world applications. This thesis makes a novel contribution by developing a user-preference-based framework for Bayesian hyperparameter optimisation across smart energy and environmental applications. The proposed approach enables systematic tuning of model hyperparameters to balance accuracy and computational cost according to application requirements. Overall, this work demonstrates how data-driven optimisation can enhance predictive performance and support the development of efficient and sustainable smart energy and environmental systems.<p></p>"],"dc:identifier":["10779/exe.31970667.v1"],"dc:relation":["https://figshare.com/articles/thesis/Bayesian_Optimisation_of_Hyperparameters_in_Regression_Models_for_Smart_Energy_and_Environmental_Systems/31970667"],"dc:rights":["All rights reserved","Open Access after 2027-10-13"],"dc:subject":["Tuned Regression","Smart Home Energy"],"dc:title":["Bayesian Optimisation of Hyperparameters in Regression Models for Smart Energy and Environmental Systems"],"dc:type":["Text","Thesis"]},"updated_at":"2026-07-27T19:33:32Z"}