Analysis and model construction of college students’ behavior portrait based on big data technology

Under the background of education informatization, the establishment of digital and intelligent smart campus platform has become an inevitable trend, aiming at realizing intelligent education, management and service. Based on big data technology, this paper applies machine learning technology to the study of student behavior portrait to build a comprehensive student data portrait. The main research focuses on building student behavior models using machine learning algorithms, especially solving the challenge that K-means algorithms are susceptible to local optimal solutions. To overcome this limitation, an improved K-means algorithm based on Canopy clustering and max-min distance principle is proposed. By using the improved K-means algorithm, cluster analysis of student data from two dimensions of academic performance and consumer behavior is carried out, and then word cloud visualization technology is used to generate student portraits. The experimental results show that the improved K-means algorithm can effectively distinguish students with different behavior characteristics, so that educational institutions can fully understand student behavior. This method helps to accurately predict student behavior and realize personalized service, which makes an important contribution to the construction of smart campus in colleges and universities. By combining big data technology and machine learning, this research aims to revolutionize student management practices and promote a more proactive and personalized development of education and campus services.