SYSTEM FOR ANALYZING LARGE DATABASES USING DECISION TREES

Introduction. The growth in computer power and the internet has led to a significant increase in the volumes of data collected by organizations and individuals, and this trend is only growing. This growth affects many sciences, changing approaches in medicine, biology, sociology, and other fields. Modern technical developments allow the use of complex algorithms, artificial neural networks, and simulation models to solve tasks that previously required an expert approach. The increase in computational power has also enabled the use of algorithms that were previously limited to scientific interest for real multidimensional tasks. Big data is increasingly penetrating small and medium-sized businesses, not only for analyzing their own statistics but also for using the data of large operators for analytical purposes. In this case, entrepreneurs face high costs of using such data, a lack of qualified specialists, and necessary computational capacities. Purpose is to develop a concept, model, construct, and implement a specialized recommendation system that uses a hybrid model. The source of data for the system includes expert knowledge and additional information about medical substances, including composition, analogs, treatment protocols, and their success rates. The project provides businesses with access to high-quality analytical tools, minimizing the need for significant investments in developing their own analytical departments. Materials and methods include machine learning and artificial neural networks; a ranking method based on user interaction sessions and the RankBoost algorithm; the gradient boosting method and its XGBoost implementation; modeling and constructing software in UML notation. Results. The paper develops a concept, constructs, and implements the proposed model of the recommendation system. After more extensive testing and calibration, as well as expanding the system's information base, it can be applied as a commercial project. Discussion. Further research proposes to focus on expanding the information base of the system's operation and increasing the number of functions. It is also advisable to introduce metrics for evaluating the quality of recommendations, which can be used in the algorithm for self-learning of the system's algorithms and for fine-tuning the parameters of these algorithms.