{"id":{"repo_id":"mcmaster","oai_identifier":"oai:macsphere.mcmaster.ca:11375/33767"},"canonical_url":"https://search.ndltd.org/etd/mcmaster/oai:macsphere.mcmaster.ca:11375/33767","repository":{"repo_id":"mcmaster","name":"McMaster University","base_url":"https://macsphere.mcmaster.ca/server/oai/request"},"display":{"title":"Data-Driven Resource Allocation under Uncertainty: A Framework Integrating Gaussian Process Regression with Budgeted Robust Optimization.","abstract":"Resource-allocation decisions arise in many engineering and business applications, where limited resources must be distributed among competing activities to maximize overall performance. In data-driven settings, these decisions are often based on estimated resource–response relationships that are subject to uncertainty arising from observation variability, limited data, and model prediction error. Ignoring this uncertainty can lead to allocations that perform well under nominal predictions but may lead to poor performance when realized responses are lower than predicted. This thesis develops an integrated framework for data-driven resource allocation under predictive uncertainty. Nonlinear resource–response relationships are first estimated from historical data, with Gaussian process regression used to provide both response predictions and associated predictive uncertainty. Monotonicity constraints are incorporated when supported by prior knowledge of the response behavior. The resulting continuous allocation problem is discretized into a mixed-integer linear formulation, allowing predictive uncertainty to be incorporated through a budgeted robust optimization model. An uncertainty budget controls the degree of protection against simultaneous adverse deviations and provides a systematic trade-off between nominal performance and robustness. The framework is demonstrated through two applications: marketing budget allocation across advertising segments and feedstock allocation among parallel ethylene cracking furnaces. The two case studies demonstrate how predictive modeling, uncertainty quantification, and robust optimization can be combined within a general decision-support framework for nonlinear resource-allocation problems.","abstract_html":"Resource-allocation decisions arise in many engineering and business applications, where limited resources must be distributed among competing activities to maximize overall performance. In data-driven settings, these decisions are often based on estimated resource–response relationships that are subject to uncertainty arising from observation variability, limited data, and model prediction error. Ignoring this uncertainty can lead to allocations that perform well under nominal predictions but may lead to poor performance when realized responses are lower than predicted. This thesis develops an integrated framework for data-driven resource allocation under predictive uncertainty. Nonlinear resource–response relationships are first estimated from historical data, with Gaussian process regression used to provide both response predictions and associated predictive uncertainty. Monotonicity constraints are incorporated when supported by prior knowledge of the response behavior. The resulting continuous allocation problem is discretized into a mixed-integer linear formulation, allowing predictive uncertainty to be incorporated through a budgeted robust optimization model. An uncertainty budget controls the degree of protection against simultaneous adverse deviations and provides a systematic trade-off between nominal performance and robustness. The framework is demonstrated through two applications: marketing budget allocation across advertising segments and feedstock allocation among parallel ethylene cracking furnaces. The two case studies demonstrate how predictive modeling, uncertainty quantification, and robust optimization can be combined within a general decision-support framework for nonlinear resource-allocation problems.","abstract_has_math":false,"creators":["He, Xiaoxue"],"institution":null,"degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":"Chemical Engineering","school":null,"contributors":[],"advisors":["Swartz, Christopher","Huang, Kai"],"committee_chairs":[],"committee_members":[],"year":2026,"date_issued":"2026","date_published":"2026","updated_at":"2026-11-18T19:33:48Z","subjects":["Robust Optimization","Resource Allocation","Data-Driven Modeling"],"languages":["en"],"rights":[],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://doi.org/10.71548/25537"],"render_values":[{"text":"https://doi.org/10.71548/25537","href":"https://doi.org/10.71548/25537","code":true}]}],"person_identifier_entries":[]},"links":{"outbound_url":"https://hdl.handle.net/11375/33767","outbound_label":"Handle","outbound_source":"dc:identifier.uri","record_url":"https://hdl.handle.net/11375/33767","record_label":"Repository record","record_kind":"handle","record_source":"dc:identifier.uri","download_url":null,"download_label":null,"download_source":null,"access_status":null,"embargo_until":null,"access_note":null},"source_record":{"url":"https://macsphere.mcmaster.ca/server/oai/request?verb=GetRecord&metadataPrefix=dim&identifier=oai%3Amacsphere.mcmaster.ca%3A11375%2F33767","prefix":"dim"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Swartz, Christopher","Huang, Kai"]},{"key":"dc:contributor.department","label":"Department","values":["Chemical Engineering"]},{"key":"dc:creator","label":"Author","values":["He, Xiaoxue"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2026-10-14T01:19:56Z"]},{"key":"dc:date.issued","label":"Date","values":["2026"]},{"key":"dc:type","label":"Dc Type","values":["Thesis"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Robust Optimization","Resource Allocation","Data-Driven Modeling"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.iso","label":"Language (ISO)","values":["en"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://hdl.handle.net/11375/33767","https://doi.org/10.71548/25537"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["Resource-allocation decisions arise in many engineering and business applications, where limited resources must be distributed among competing activities to maximize overall performance. In data-driven settings, these decisions are often based on estimated resource–response relationships that are subject to uncertainty arising from observation variability, limited data, and model prediction error. Ignoring this uncertainty can lead to allocations that perform well under nominal predictions but may lead to poor performance when realized responses are lower than predicted. This thesis develops an integrated framework for data-driven resource allocation under predictive uncertainty. Nonlinear resource–response relationships are first estimated from historical data, with Gaussian process regression used to provide both response predictions and associated predictive uncertainty. Monotonicity constraints are incorporated when supported by prior knowledge of the response behavior. The resulting continuous allocation problem is discretized into a mixed-integer linear formulation, allowing predictive uncertainty to be incorporated through a budgeted robust optimization model. An uncertainty budget controls the degree of protection against simultaneous adverse deviations and provides a systematic trade-off between nominal performance and robustness. The framework is demonstrated through two applications: marketing budget allocation across advertising segments and feedstock allocation among parallel ethylene cracking furnaces. The two case studies demonstrate how predictive modeling, uncertainty quantification, and robust optimization can be combined within a general decision-support framework for nonlinear resource-allocation problems."]},{"key":"dc:description.degree","label":"Dc Description Degree","values":["Master of Applied Science (MASc)"]},{"key":"dc:title","label":"Title","values":["Data-Driven Resource Allocation under Uncertainty: A Framework Integrating Gaussian Process Regression with Budgeted Robust Optimization."]},{"key":"others:access_status","label":"Others Access Status","values":["embargo"]}]}],"canonical_facts":{"dc:contributor.advisor":["Swartz, Christopher","Huang, Kai"],"dc:contributor.department":["Chemical Engineering"],"dc:creator":["He, Xiaoxue"],"dc:date.accessioned":["2026-10-14T01:19:56Z"],"dc:date.issued":["2026"],"dc:description.abstract":["Resource-allocation decisions arise in many engineering and business applications, where limited resources must be distributed among competing activities to maximize overall performance. In data-driven settings, these decisions are often based on estimated resource–response relationships that are subject to uncertainty arising from observation variability, limited data, and model prediction error. Ignoring this uncertainty can lead to allocations that perform well under nominal predictions but may lead to poor performance when realized responses are lower than predicted. This thesis develops an integrated framework for data-driven resource allocation under predictive uncertainty. Nonlinear resource–response relationships are first estimated from historical data, with Gaussian process regression used to provide both response predictions and associated predictive uncertainty. Monotonicity constraints are incorporated when supported by prior knowledge of the response behavior. The resulting continuous allocation problem is discretized into a mixed-integer linear formulation, allowing predictive uncertainty to be incorporated through a budgeted robust optimization model. An uncertainty budget controls the degree of protection against simultaneous adverse deviations and provides a systematic trade-off between nominal performance and robustness. The framework is demonstrated through two applications: marketing budget allocation across advertising segments and feedstock allocation among parallel ethylene cracking furnaces. The two case studies demonstrate how predictive modeling, uncertainty quantification, and robust optimization can be combined within a general decision-support framework for nonlinear resource-allocation problems."],"dc:description.degree":["Master of Applied Science (MASc)"],"dc:identifier.uri":["https://hdl.handle.net/11375/33767","https://doi.org/10.71548/25537"],"dc:language.iso":["en"],"dc:subject":["Robust Optimization","Resource Allocation","Data-Driven Modeling"],"dc:title":["Data-Driven Resource Allocation under Uncertainty: A Framework Integrating Gaussian Process Regression with Budgeted Robust Optimization."],"dc:type":["Thesis"],"others:access_status":["embargo"]},"updated_at":"2026-11-18T19:33:48Z"}