Customer-driven Prescriptive Analytics for Logistics Planning

CUSTOMIZE asked whether transaction data could be used to predict demand at this finer resolution, and whether those predictions would change how resources should be positioned. It developed methods to infer customer-level demand patterns from transaction records and to enrich these predictions with publicly available data. These forecasts were integrated into prescriptive models that determined how many resources were needed and where they should be positioned as demand evolved. The approach was studied in two contrasting settings: on-demand food delivery and home healthcare logistics.