Gerardo Berbeglia is an Associate Professor at Melbourne Business School (University of Melbourne). He has a PhD in Operations Research from the Universit\u00e9 de Montr\u00e9al, and a Master’s degree in Computer Science from the University of Buenos Aires. Prior to joining MBS, Gerardo was a Senior Scientist at ExPretio Technologies Inc, and, later, a postdoctoral fellow at McGill University.<\/p>\n
<\/p>\n
Gerardo\u2019s research has been published in leading journals and conference proceedings, including Management Science, Transportation Science, INFORMS Journal on Computing, European Journal of Operational Research, Operations Research Letters, Algorithmica, Journal of Mathematical and Economics, ACM conference on Economics and Computation (EC), Conference on Web and Internet Economics (WINE) and the International Joint Conference on Artificial Intelligence.<\/em> His recent research focuses on revenue management, choice modelling, and quantitative models that account for social influence in online markets.<\/p>\n <\/p>\n Gerardo teaches Operations, Optimisation and Decision Making, and Supply Chain Analytics in the MBA and Master of Business Analytics programs. He received the MBS Annual Teaching Excellence Award five times.<\/p>\n"}],"event_agenda":false,"event_photo_gallery":false,"event_presentations":false,"event_custom_heading":[{"event_custom_title":"Abstract","event_custom_details":" Traditional assortment optimisation aims to increase profits by strategically removing certain products, causing some of the lost demand to shift towards higher-margin alternatives. In contrast, refined assortment optimisation takes a more nuanced approach: instead of making products unavailable, it selectively reduces their appeal to steer customers towards more profitable options. Techniques for diminishing product attractiveness include extending delivery times, requiring advance purchases, imposing waiting lists, and reducing product features.<\/p>\n <\/p>\n Refined assortment optimisation can substantially increase expected profits over traditional methods. For the Random Consideration Set choice models, profits under refined assortment optimisation can double those obtained using traditional assortment optimisation. For the latent-class Multinomial Logit (MNL) model, the profit improvement ratio can reach up to the minimum of the number of products and the number of market segments. However, this optimisation is computationally intractable, necessitating effective heuristics. We demonstrate that the revenue-ordered heuristic maintains the same performance guarantee for the personalised refined assortment optimisation problem as it does for the traditional non-personalised problem. Building on this insight, we develop revenue-ordered heuristics tailored specifically for the refined assortment problem.<\/p>\n <\/p>\n Our experiments using the latent-class MNL model show that these heuristics often yield expected revenues exceeding those of the optimal solutions to the traditional assortment optimisation problem. Extensive numerical studies further reveal that our heuristics outperform both the adapted ADXOPT algorithm for refined assortments and a general-purpose nonlinear solver.<\/p>\n <\/p>\n Joint work with Alvaro Flores and Guillermo Gallego<\/p>\n"}],"event_enquiry_details":{"event_enq_full_name":"","event_enq_department":"","event_enq_email":"","event_enq_telephone":"","event_enq_website":""}}]