Final Report Summary - FLEXREV (Integrating Flexible Discrete Choice and Revenue Management Models) The research objective in this project is to design new revenue management (RM) systems that integrate consumer choices to the mathematical models of RM. The research is comprised of two main sub-thrusts: 1) Estimation of Network Generalized Extreme Value (NetGEV) models for choice-based samples 2) Integration of choice based models with capacity based RM models. In addition to these research objectives, the project’s mobility and integration objectives are to promote the reintegration of the researcher to her country and host institution, and to establish collaborations with local and European researches. In order to fulfill these objectives, a vast amount of work has been carried out, and the results achieved are summarized below:NetGEV model shows a very high potential for accurately modeling constrained demand as the collection of individual consumers’ purchases as well as for understanding demand shifts caused when products become unavailable. Two different NetGEV models, Multinomial Logit (MNL) and Nested Logit (NL) have been developed to model consumer behavior. Using MNL, first choice based models have been developed which covers extensions with time varying demand and time varying choice sets. In addition, an estimation method has been developed to efficiently estimate the model parameters while integrating the no-purchase alternative. The model was much more efficient and stable compared to counterparts in the literature. All these work has been done for both single and multi-product environments. In addition, classical Expected Marginal Seat Revenue methodology has also been updated for multi-product, multi-capacity environment. To test the developed methods, data from two different major international hotel chains has been used. Data requirements, collection process and data analysis with purpose of using these with choice based models have been discussed. Parameter estimates have been found using developed methods. With these results, simulation studies have been performed in order to allocate resources to customers to maximize revenue. Several allocation algorithms for several estimation methods have been developed as well. These algorithms were tested under different scenarios to observe robustness of these algorithms and quality of estimation methods.In order to use NL models, the data was structured to be categorized into the nests. A network tree with two levels has been constructed. Data has been grouped in several ways to find the best fit. Simultaneous estimation has been used as part of the project. As an addition, customer churn has been modeled with NL at telecommunication industry, using real data from a major firm. Both of these showed promising results for usage of NL models to estimate consumer choice behavior. Regarding the fulfillment of integration objectives, the researcher was an invited speaker in a seminar series in Bilkent University. In addition, the researcher advised a Master’s thesis on modeling consumer behavior with NL models at telecommunication sector’s churn behavior. The results of this thesis have been published at a peer-reviewed and SSCI journal. Also, the researcher has advised another Master’s thesis at TOBB ETU, whose topic was on consumer allocation using several different allocation methods including choice models at hotels. Data from a local but big hotel chain was used in this study. A paper out of the results of this thesis has been accepted for publication at a peer-reviewed journal. In addition the researcher attended the 2014 INFORMS Revenue Management and Pricing Section Conference. These activities resulted in fruitful collaborations from local, European and other international researchers. Currently, the researcher is advising 2 more Master’s students and their work include RM applications at low-cost airlines, as well as other RM issues at hotels. They are planning to incorporate results from this project into their new studies.