4148. Integrating Predictive and Prescriptive Analytics for Assortment Optimization - a Machine Learning-based Approach
Invited abstract in session MC-59: Pricing and learning 2, stream Pricing and Revenue Management.
Monday, 12:30-14:00Room: S08 (building: 101)
Authors (first author is the speaker)
| 1. | Niloufar Sadeghi
|
| Chair of Service Operations Management, University of Mannheim | |
| 2. | Siamak Khayyati
|
| HEC Liege, University of Liege | |
| 3. | Cornelia Schoen
|
| Chair of Service Operations Management, University of Mannheim |
Abstract
Product line design is one of the key problems that firms need to solve. Therefore, they employ techniques to predict customer choice behaviour and optimize the performance of the product assortment they aim to offer.In most cases, the prediction problem and the optimization problem are defined as separate problems and solved in a sequential manner where first, the choice model is estimated and second, the assortment optimization problem is solved, using the choice model parameters as an input. Integrating estimation and optimization provides an opportunity for the empirical model to be more accurate where it matters – close to the optimal assortment. We develop a MILP formulation that is able to solve the two problems jointly. Using numerical experiments, we analyse under what conditions and to which extent the integrated approach is superior to the sequential approach.
Keywords
- Revenue Management and Pricing
Status: accepted
Back to the list of papers