A Data-Driven Approach to Constraint Optimization

Many problems occurring in production, transport, supply chains and everyday life problems can be formulated in the form of constraint optimization problems (COPs). Most often these are issues related to planning and scheduling, distribution of resources, fleet selection, route and network optimization, configuration of machines and manufacturing systems, timetabling, etc. In the vast majority of cases, these are discrete problems of a combinatorial nature. Significant difficulties in modelling and solving COPs are usually the magnitude of real problems, which translates into a large number of variables and constraints as well as high computational complexity (usually NP-hard problems). The article proposes a data-driven approach, which allows a significant reduction in the magnitude of modelled problems and, consequently, the possibility of solving many real problems in an acceptable time.