Development and Comparison of Models for Forecasting the Demand for Goods and Services Based on User's Purchases

This paper presents an overview of the ways to use recommendation systems in the field of video content recommendations on online platforms and provides examples of the use of recommendation algorithms on world-famous websites with millions of users every day. Each of the platforms has its own specifics of work, its advantages and disadvantages, but they are united by the fact that in order to provide high-quality recommendations, sites process the user's browsing history for the entire time of their presence on the platform, as well as take into account personal data. In connection with this, it was decided to develop an algorithm for a session recommendation system, for which it was necessary to collect a large amount of training data from various sources. The paper presents the results of the obtained model and conclusions about the possibility of developing and applying the model. The work consists of 47 pages, contains 14 illustrations, uses 26 sources.