The personalized recommendation of public art education aims to provide art education resources and services that better meet the needs of students according to their individual needs and interests. The personalized recommendation system of public art education in higher education has serious error rate. The K-means algorithm is an algorithm that focuses on personalized recommendations. This paper uses K-means algorithm to establish a personalized recommendation system for public art education in higher education and draws conclusions through experiments. The algorithm greatly reduces the error rate, and its accuracy rate reaches 97.8%, and greatly improves the efficiency of the personalized recommendation system for public art education in higher education.