AutoML Approach to Recommender System

This paper focuses on the relevant topic of automation in the sphere of machine learning: how to build a universal recommender system. Such automated machine learning systems are highly demanded in industry due to the presence of large amount of routine data processing operations as well as testing various combinations of parameters. In this paper, an automated recommender system is proposed and demonstrated. Thus, the paper describes various types of recommender systems, with an emphasis on collaborative filtering approaches. Different methods for machine learning automation are considered to be incorporated into the recommender system under development. The effectiveness of the proposed approach is judged by comparing the output with and without the AutoML system. The result of this paper is a comparative analysis of the developed and proposed algorithms (systems); it is shown that the recommender system with AutoML provides better quality indicators than the system created without the proposed approach.