Facing the huge competition pressure of modern generation, it is of practical significance to evaluate the comprehensive abilitiy of college students. Objective evaluation can make both college students and teachers realize the shortcomings of students' development and make progress. However, in reality, many colleges lack an objective evaluation system and rely more on subjective judgments of personal emotions. Therefore, this paper wants to establish an objective and fair evaluation system for the comprehensive development level of college students, and proposes the use of machine learning models for adaptation to provide support for scientific evaluation of college students' comprehensive ability.