In light of this scenario, Lumini was challenged to evaluate the dataset of all delivery routes carried out, Starting the construction of an algorithm for route optimization and suggestions for improvements based on statistics.Thus, the IT company began the project by analyzing the data and information regarding the delivery routes that year. We divided the project into three stages: first, we performed data treatment and cleaning, then exploratory data analysis, and finally, the generation of propensity models. We analyzed the cities with the highest number of deliveries. “We analyzed the activities by distribution center, noting that one of them was responsible for more than 50% of the deliveries.” The solution adopted was based on the relationship between the cities in the routes taken. Based on the correlations found among them, it was possible to verify whether the cities considered important were being served by the best distribution centers. Once the relevant routes were filtered, the next step was to evaluate them. To do this, the cities on a route were ordered so that the distance traveled would be as short as possible. To create these optimized routes, the following assumptions were made: the truck cannot pass through the same city twice on the route, must complete 100% deliveries, and must return to the starting point. Thus, Lumini calculated the total distance traveled and the optimal starting point (considering the available distribution centers). The total distance traveled was used to compare the current routes with the optimized routes.