Effective entrepreneurship education can not only cultivate students’ entrepreneurship awareness and inspire entrepreneurial potential, but also set up an entrepreneurial foundation and form entrepreneurial practice, but the current teaching system doesn’t have strong timeliness, due to the high interest of students and lack of rich teaching methods and means by college teachers. Taking “online + offline” entrepreneurship education courses as an example, an “online + offline” teaching mode based on case teaching method was developed in this paper. Based on curriculum theory, the author identified the attractive quality and must-be quality of this course by fully understanding students’ needs in the learning process of “online + offline” courses, and designed a questionnaire on the needs for the learning support services of “online + offline” course products, to understand the satisfaction degree of students corresponding to each need. At the same time, combined with the characteristics of the “online + offline” courses, by reference to the evaluation criteria of DEMATEL-ANP, course mentoring, communication, final exam, subtitles and video effect were taken as key indicators, to build an evaluation system to be used in this mode. Finally, the teaching practice proves that this mode can better increase students’ learning interest, expand the teaching content of entrepreneurship education course and improve students’ satisfaction with this course.

The current traditional college psychology courses only pay attention to the prevention and correction of the psychological problems of college students, but neglect to let students have a good positive emotion to learn, which is not conducive to the development of psychological potential of students, thus psychology learning cannot achieve the expected results. Based on the psychology blended teaching theory, combined with the entrepreneurial education of college students, a set of PDCA (Plan-Do-Check-Action) teaching mode which adapts to the development needs of "Internet education" is designed, and formed several new methods of teaching, such as "integration of online and offline classroom teaching". In addition, a new performance prediction model based on graph convolution neural network is proposed decision tree data mining methods are used to predict student performance, by extracting user information and high-dimensional information of the knowledge graph, this mode is conducive to platform managers to understand students’ conditions in time and improve teaching quality. This teaching mode dynamically analyzes the internal relationships among learners, teachers, online classrooms, offline classrooms and it can find problems early and provide solutions. It is conducive to improving students' academic performance and learning enthusiasm, and enhancing the ability of college students.