Logistics problems require special attention, because every year they become more complicated and multivariable. On the one hand, a supply chain management includes
incessant monitoring of such issues as requests elaboration, paths determination, routing of shipments, multimodal choice, set up of transhipments, fleet choice and maintenance, warehousing, packaging and others. On the other hand, dozens of people are involved in the logistics process. All these moments complicate the decision-making that is why data driven decisions are required nowadays. As well as shipment problems are NP-hard, the heuristic methods should be applied to resolve them. In this article we propose a genetic algorithm to solve the complex problem that consists of the Travelling Salesman Problem combined with the Knapsack Problem. We have developed an urban freight transportation model which is focused on the minimization of the underway time as well as on the maximization of the truck’s loading. A significant contribution in our method is the census of traffic frequency by using traffic zoning. The developed approach has been implemented using the Python programming language in the Zeppelin environment. The first version of the system has been approved in the city of Samara (Russia) with test demand dataset.