A framework for inherently interpretable optimization models

With dramatic improvements in optimization software, the solution of large-scale problems that seemed intractable decades ago are now a routine task. This puts even more real-world applications into reach optimizers. At same time, solving often turns out to be one smaller difficulties when putting solutions practice. One major barrier is software can perceived as black box, which may produce high quality, but create completely different circumstances change leading low acceptance optimized solutions. Such issues interpretability and explainability have seen significant attention other areas, such machine learning, less so optimization. In this paper we propose an framework inherently comes with easily interpretable rule, explains under certain chosen. Focusing on univariate decision trees represent rules, integer programming formulations well heuristic method ensure applicability our approach for problems. By presenting several extensions tree approach, showcase generality proposed framework. Computational experiments using random data road network indicate costs inherent very small.