Grey relational analysis model of classified variables and its application in Jiangsu universities' entrepreneurship education achievements

摘要

Nowadays, entrepreneurship is drawing more and more attention all over China. Many policies are carried out to promote entrepreneurship and innovation. As the vital implementation subjects of entrepreneurship education, universities are providing strong fiscal, policy support and human resources for students' entrepreneurship practices, launching courses on entrepreneurship as well as encouraging students to participate in various entrepreneurship competitions. Jiangsu, as an educational-clustered province with the largest numbers of university and key laboratory is actively echoing with the higher education trend. With the approach of “Popular Entrepreneurship and Innovation” era, Jiangsu universities are responsible to boost the development of New Business Initiative Plan. The current Grey Relational Analysis (hereafter as GRA) model hasn't involved classified variables yet, while many variables like age, gender are shown as classified variables in the field of social science and economics. Thus, this thesis attempts to set a GRA model of classified variables for the sake of practical analysis. Statistics and Econometrics researches always use virtual variables to represent classified variables. Nevertheless, applying virtual variables may cause some problems in GRA. Because of the application of virtual variables, the area between the classified variables sequence and the X axis is relatively fixed, which will have impact on the calculation results of GRA. In addition, the results will be sensible to the sequence dimension. We divide the sequence into several subsequences based on the category. If the classified variables have no influence on each other, there will be no distance between the subsequences. In this way, we convert the relationship of classified variables to that of subsequences. Then, the idea of absolute degree of grey incidence can be applied to calculate the GRA of subsequences and describe the effects of classified variables. During the calculation, the sequence is divided into several subsequences based on classified variables. Then, the equilong subsequences are constructed. Normally, the lengths of subsequences can hardly become equal, so we firstly sort the subsequences by the shortest sequence length. According to the Most Similar Principle, we cut out and generate all the subsequences. In this way, we achieve the equilong subsequences. Finally, the GRA is calculated out. As the case study, we use the above-mentioned GRA model to study whether colleges/universities types have any impacts on the effects of university entrepreneurship education. Based on the goal-oriented, scientific, complete, systematic and operability principles of the evaluation index, the determined evaluation indexes which reflect the effect of university Innovation and Entrepreneurship Education(hereafter as IEE) input include entrepreneurship project approval, ratio of self-employed students, and Entrepreneurship Competition Awards. In the method of questionnaires and interviews, the research investigates 150 relevant departments, such as the Department of Student Affairs, the Office of Academic Affairs and the Innovation Incubation Center of 36 universities in Jiangsu. To ensure the comprehensiveness and general applicability, the respondents cover 985 and 211 Project universities, ordinary universities as well as vocational colleges. One hundred and eight among 150 questionnaires have been taken back for providing the effective data. The valid withdrawal ratio is 72%. According to the calculation method, we calculated the GRA between all the indexes of IEE achievements and the colleges/universities types. The results show that colleges/universities type has the greatest relevancy with the Entrepreneurship Competition Awards, which is followed by the number of Entrepreneurship Project Approval; the Ratio of Self-employed Students is the least relevant with colleges/universities type. The larger the relevancy is, the more the