2020/08/04 by Miguel Ángel Muñoz, Muñoz, Miguel Angel, Salvador Pineda +3 · 1 citation
Decision Sciences · Engineering · #FOS: Mathematics #Optimization and Control (math.OC) #Probabilistic and Robust Engineering Design #Reservoir Engineering and Simulation Methods #Risk and Portfolio Optimization
paper · pdf · doi:10.48550/arxiv.2008.01500
openalex publication_date 2020/08/04 · openalex created_date 2022/07/26 · openalex updated_date 2026/07/28
In this paper, we propose a novel approach for data-driven decision-making\nunder uncertainty in the presence of contextual information. Given a finite\ncollection of observations of the uncertain parameters and potential\nexplanatory variables (i.e., the contextual information), our approach fits a\nparametric model to those data that is specifically tailored to maximizing the\ndecision value, while accounting for possible feasibility constraints. From a\nmathematical point of view, our framework translates into a bilevel program,\nfor which we provide both a fast regularization procedure and a big-M-based\nreformulation that can be solved using off-the-shelf optimization solvers. We\nshowcase the benefits of moving from the traditional scheme for model\nestimation (based on statistical quality metrics) to decision-guided prediction\nusing three different practical problems. We also compare our approach with\nexisting ones in a realistic case study that considers a strategic power\nproducer that participates in the Iberian electricity market. Finally, we use\nthese numerical simulations to analyze the conditions (in terms of the firm's\ncost structure and production capacity) under which our approach proves to be\nmore advantageous to the producer.\n