{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/from-predictive-to-prescriptive-analytics","title":"From Predictive to Prescriptive Analytics","arxiv_id":"1402.5481","date":"2014-02-22","proceeding":null,"authors":["Dimitris Bertsimas","Nathan Kallus"],"abstract":"In this paper, we combine ideas from machine learning (ML) and operations\nresearch and management science (OR/MS) in developing a framework, along with\nspecific methods, for using data to prescribe optimal decisions in OR/MS\nproblems. In a departure from other work on data-driven optimization and\nreflecting our practical experience with the data available in applications of\nOR/MS, we consider data consisting, not only of observations of quantities with\ndirect effect on costs/revenues, such as demand or returns, but predominantly\nof observations of associated auxiliary quantities. The main problem of\ninterest is a conditional stochastic optimization problem, given imperfect\nobservations, where the joint probability distributions that specify the\nproblem are unknown. We demonstrate that our proposed solution methods, which\nare inspired by ML methods such as local regression, CART, and random forests,\nare generally applicable to a wide range of decision problems. We prove that\nthey are tractable and asymptotically optimal even when data is not iid and may\nbe censored. We extend this to the case where decision variables may directly\naffect uncertainty in unknown ways, such as pricing's effect on demand. As an\nanalogue to R^2, we develop a metric P termed the coefficient of\nprescriptiveness to measure the prescriptive content of data and the efficacy\nof a policy from an operations perspective. To demonstrate the power of our\napproach in a real-world setting we study an inventory management problem faced\nby the distribution arm of an international media conglomerate, which ships an\naverage of 1bil units per year. We leverage internal data and public online\ndata harvested from IMDb, Rotten Tomatoes, and Google to prescribe operational\ndecisions that outperform baseline measures. Specifically, the data we collect,\nleveraged by our methods, accounts for an 88\\% improvement as measured by our\nP.","url_abs":"http://arxiv.org/abs/1402.5481v4","url_pdf":"http://arxiv.org/pdf/1402.5481v4.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"from-predictive-to-prescriptive-analytics","repo_url":"https://github.com/yf-yang/Portfolio-Optimization","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}}],"tasks":[{"task_slug":"management","task_name":"Management"},{"task_slug":"stochastic-optimization","task_name":"Stochastic Optimization"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1402.5481","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}