{"id":{"repo_id":"uiuc","oai_identifier":"oai:www.ideals.illinois.edu:2142/117593"},"canonical_url":"https://search.dev.ndltd.org/etd/uiuc/oai:www.ideals.illinois.edu:2142/117593","repository":{"repo_id":"uiuc","name":"University of Illinois - Urbana-Champaign","base_url":"https://www.ideals.illinois.edu/oai-pmh"},"display":{"title":"Dynamic pricing of airline ancillaries under distribution shift","abstract":"Submission published under a 24 month embargo labeled 'Closed Access', the embargo will last until 2024-12-01","abstract_html":"Submission published under a 24 month embargo labeled 'Closed Access', the embargo will last until 2024-12-01","abstract_has_math":false,"creators":["Garg, Abhinav"],"institution":"University of Illinois at Urbana-Champaign","degree_name":"M.S.","degree_level":"Thesis","degree_discipline":"Industrial Engineering","degree_department":null,"school":null,"contributors":["Marla, Lavanya"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2022,"date_issued":"2022-12","date_published":"2022-12","updated_at":"2026-07-22T22:24:56Z","subjects":["Machine Learning","Dynamic Pricing","Airline Ancillaries","Distribution Shift","Causal Inference","Covariate Shift","Concept Drift"],"languages":["en","eng"],"rights":["Copyright 2022 Abhinav Garg"],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"https://hdl.handle.net/2142/117593","outbound_label":"Handle","outbound_source":"dc:identifier"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Marla, Lavanya"]},{"key":"dc:creator","label":"Author","values":["Garg, Abhinav"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2022-12","2022-12-05"]},{"key":"dc:type","label":"Dc Type","values":["text","Thesis"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Industrial Engineering"]},{"key":"thesis:degree_level","label":"Degree Level","values":["Thesis"]},{"key":"thesis:degree_name","label":"Degree Name","values":["M.S."]},{"key":"thesis:institution_name","label":"Thesis Institution Name","values":["University of Illinois at Urbana-Champaign"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Machine Learning","Dynamic Pricing","Airline Ancillaries","Distribution Shift","Causal Inference","Covariate Shift","Concept Drift"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["en","eng"]},{"key":"dc:rights","label":"Dc Rights","values":["Copyright 2022 Abhinav Garg"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["https://hdl.handle.net/2142/117593"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["Submission published under a 24 month embargo labeled 'Closed Access', the embargo will last until 2024-12-01","The student, Abhinav Garg, accepted the attached license on 2022-12-01 at 16:40.","The student, Abhinav Garg, submitted this Thesis for approval on 2022-12-01 at 17:03.","This Thesis was approved for publication on 2022-12-05 at 15:27.","DSpace SAF Submission Ingestion Package generated from Vireo submission #18726 on 2023-04-12 at 11:36:45","Many machine learning algorithms rely on the assumption that the test data are drawn from the same distribution as the training data to guarantee performance. In real-world applications, however, this assumption may be violated due to no prior knowledge on the test data and ever changing customer behavior in a dynamic environment. These violations may become even more intense in a decision-making scenario when the system experiences unanticipated shocks such as COVID-19. One such application is personalized contextual pricing of airline ancillaries since COVID-19 impacted travel demand in a way that was never seen before because of the speed and scale at which the demand dropped, as well as the rapidly changing customer purchase behavior. In this study, we explore the problem of distribution shift in the form of covariate shift and concept drift for the airline ancillary purchase use-case. We present methods for detecting such shifts in the customer behavior created by COVID-19 and use this new information to build robust pricing models to find optimal price for each ancillary for each customer that he/she is willing to pay. Recently, subset scanning techniques have emerged that use bayesian modeling to find anomalous patterns in a data set. Using such techniques, we present a framework to detect covariate shift in time when the ground truth information is not available and pinpoint attributes of customer that show different travel behavior during COVID-19. Additionally, causal forests are used to measure heterogeneous treatment effect created by an intervention in the system. We use causal forests to detect concept drift that can provide important information for building machine learning models that are robust to distribution shifts. We present and compare two methods for contextual dynamic pricing of ancillaries utilizing causal information in the form of: (1) probability calibration of a two-stage forecasting and optimization model using a logistic mapping function; (2) tuning a single-stage end-to-end deep neural network that recommends an optimal ancillary price by modeling the customer’s willingness to pay. Extensive experiments on real-world airline customer data show that our proposed methods can improve the pricing model performance and stabilize the prediction under distribution shift through simulations in an offline setting."]},{"key":"dc:format","label":"Dc Format","values":["application/pdf"]},{"key":"dc:title","label":"Title","values":["Dynamic pricing of airline ancillaries under distribution shift"]}]}],"canonical_facts":{"dc:contributor":["Marla, Lavanya"],"dc:creator":["Garg, Abhinav"],"dc:date":["2022-12","2022-12-05"],"dc:description":["Submission published under a 24 month embargo labeled 'Closed Access', the embargo will last until 2024-12-01","The student, Abhinav Garg, accepted the attached license on 2022-12-01 at 16:40.","The student, Abhinav Garg, submitted this Thesis for approval on 2022-12-01 at 17:03.","This Thesis was approved for publication on 2022-12-05 at 15:27.","DSpace SAF Submission Ingestion Package generated from Vireo submission #18726 on 2023-04-12 at 11:36:45","Many machine learning algorithms rely on the assumption that the test data are drawn from the same distribution as the training data to guarantee performance. In real-world applications, however, this assumption may be violated due to no prior knowledge on the test data and ever changing customer behavior in a dynamic environment. These violations may become even more intense in a decision-making scenario when the system experiences unanticipated shocks such as COVID-19. One such application is personalized contextual pricing of airline ancillaries since COVID-19 impacted travel demand in a way that was never seen before because of the speed and scale at which the demand dropped, as well as the rapidly changing customer purchase behavior. In this study, we explore the problem of distribution shift in the form of covariate shift and concept drift for the airline ancillary purchase use-case. We present methods for detecting such shifts in the customer behavior created by COVID-19 and use this new information to build robust pricing models to find optimal price for each ancillary for each customer that he/she is willing to pay. Recently, subset scanning techniques have emerged that use bayesian modeling to find anomalous patterns in a data set. Using such techniques, we present a framework to detect covariate shift in time when the ground truth information is not available and pinpoint attributes of customer that show different travel behavior during COVID-19. Additionally, causal forests are used to measure heterogeneous treatment effect created by an intervention in the system. We use causal forests to detect concept drift that can provide important information for building machine learning models that are robust to distribution shifts. We present and compare two methods for contextual dynamic pricing of ancillaries utilizing causal information in the form of: (1) probability calibration of a two-stage forecasting and optimization model using a logistic mapping function; (2) tuning a single-stage end-to-end deep neural network that recommends an optimal ancillary price by modeling the customer’s willingness to pay. Extensive experiments on real-world airline customer data show that our proposed methods can improve the pricing model performance and stabilize the prediction under distribution shift through simulations in an offline setting."],"dc:format":["application/pdf"],"dc:identifier":["https://hdl.handle.net/2142/117593"],"dc:language":["en","eng"],"dc:rights":["Copyright 2022 Abhinav Garg"],"dc:subject":["Machine Learning","Dynamic Pricing","Airline Ancillaries","Distribution Shift","Causal Inference","Covariate Shift","Concept Drift"],"dc:title":["Dynamic pricing of airline ancillaries under distribution shift"],"dc:type":["text","Thesis"],"thesis:degree_discipline":["Industrial Engineering"],"thesis:degree_level":["Thesis"],"thesis:degree_name":["M.S."],"thesis:institution_name":["University of Illinois at Urbana-Champaign"]},"updated_at":"2026-07-22T22:24:56Z"}