Tourist Route Optimization Using Machine Learning

The main theme of this paper is the creating of routes which are not the shortest but contains a lot of sights instead. The purpose of the paper is the identification of most effective algorithm that can build optimal tourist routes considering the number of sights on the way. To achieve it, it was necessary to solve a few tasks. 1. Collecting the data base from OpenStreetMap. 2. Making preprocessing of the data. 3. Testing optimization algorithms and identification of most effective. Algorithms which were analyzed in this paper are the Nearest neighbor algorithm, Ant colony optimization algorithm and Genetic algorithm. Testing of them provide next results. Firstly, the Nearest neighbor algorithm is more stable than others and take a little time to find an optimal route. However, the quality of paths is not as good as could be. Secondly, Ant colony optimization algorithm is best for the small samples of data and provide the most effective solutions. Thirdly, Genetic algorithm is best for the big samples of data. Due to properties of the algorithm, it can determine the number of necessary sights by itself. But if preprocessing would include the cutting of sight’s list, the Ant colony optimization algorithm also would be effective for big samples. Considering that the ACO’s time of finding the best path is smaller than Genetic algorithm’s time, ACO is the most effective algorithm. The practical significance of this paper in the potential of the results which could be used to make an app or Internet resource.