Research On The Application Of Data Mining Technology In The Regulating Matriculation For Postgraduate

Graduate regulate is an important part in the graduate recruitment progress, it is not only related to the interests of candidates, but also to the successful recruiting of the institutions enrollment. Due to the complexity and diversity of the candidates and their respective characteristics of admission units, the regulate is a kind of very complicated work. Currently, graduate regulate mainly adopts the method that the candidates analyze the characteristic of school and their own conditions, and subjectively report their voluntary transfers. However, due to the large number of admission units and their complex properties, it is very difficult for candidates to accurately analyze the characteristics of each school, which results in the a lot of unsuccessful transfers, leads to the candidates'further education failure as well as makes admissions units be unable to complete the enrollment plan. This paper applies data mining technology to into graduate regulate, provides the majority of candidates and admission units decision support, which improves the efficiency of graduate regulate, and supplies a better solution to difficult problems in graduate regulate.The ideas and work of this paper are as follows: by analyzing the characteristics of it, the graduate regulate is divided into two parts, first, make a classification of regulate schools, then schools with the similar enrollment conditions are classified as the same type, which results that school are divided into four levels; second, make a classification of the candidates in terms of their various conditions, search the type of schools which are suited to the candidate, and candidates can choose their favorite one.As for the classification of schools, by analyzing the characteristics of the school property, we can choose ID3 decision tree algorithm. According to the analysis of ID3 algorithm, we find that ID3 classification algorithm is less efficient in dealing with the classification of higher value operation, in view of this shortcoming, this paper intends to eliminate the correlation function in original formula using MacLaurin formula, resulting in improving operational efficiency. At the same time, multiply reciprocal number of the property value after the improved information entropy, thereby eliminating the affection of property value to the information entropy, avoiding calculating a larger information entropy because of some less important attributes. Compared with the traditional ID3 algorithm, the improved one tree is more feasible and efficient in establishing decision tree and making classification of schools. As to the classification of candidates, in consideration of candidates attribute's diversity, fuzziness and difficult to identify, the paper designs a semi-supervised learning algorithm which mainly has the advantage that it can classify a large number of difficult-identified candidates, using a small number of easy-identified samples. Finally, establish graduate regulate system model based improved ID3 algorithm and semi-supervised learning algorithm, and implement a simple regulate system based on the established graduate regulate system model design, the system's core features include: under the conditions recommend the appropriate level of candidate schools according to candidates conditions, school information inquiry and so on. Make a test based on the test data, the test results are basically consistent with actual regulate results, the model is feasible and has a certain value to use and promote.