Research on the Application of Data Mining Technology in the Regulating Matriculation for Postgraduate
|School||Jiangsu University of Science and Technology|
|Course||Applied Computer Technology|
|Keywords||Graduate swap Data Mining Decision Tree Semi-supervised learning|
The swap is an important part of the graduate enrollment, the ability to do the swap work not only related to the interests of the candidates, but also to the admissions unit enrollment success. Due to the respective characteristics of the candidates and the admissions unit complexity, diversity, and so do the swap work is a very complex thing. Currently, the main method of graduate swap Candidates analysis of school characteristics and conditions, subjective reporting swap volunteering. However, due to the large number of admissions unit complex attributes, candidates difficult to accurately analyze the characteristics of each school, which resulted in a lot of spice unsuccessful phenomenon leading candidates to get the job and can not complete the admissions unit enrollment plan. Data mining techniques to work of graduate swap, for the majority of the candidates and the admissions unit provides decision support to improve the efficiency of the Graduate toner, you can the better solution graduate swap work difficult problems. This paper ideas and work as follows: by analyzing the characteristics of Graduate swap, the swap is divided into two parts, the first swap school classification schools with similar enrollment conditions classified as a class, so the school is divided into four grades; followed by the candidates all aspects of the conditions of the candidates need to swap classified to identify the type of school suitable for the Candidates conditions in such schools, candidates can choose their favorite school. For the classification of the school, through the analysis of the characteristics of the school property, the choice of the ID3 algorithm, the ID3 algorithm analysis found that the ID3 algorithm is less efficient when computing the value of more processing classification results, in view of such shortcomings of this paper McLaurin formula eliminating the correlation function in the original information entropy formula, thereby improving the operation efficiency, while the calculation of information entropy for appropriate transformation, which can eliminate the influence of the attribute number of values ??of the information entropy avoid unimportant attributes calculated value more information entropy. Improved ID3 algorithm to build a decision tree, to classify the school, compared with traditional ID3 algorithm, prove the feasibility and efficiency of the improved algorithm. For candidates classified in view of the diversity of the attributes of candidates, fuzzy and difficult to identify the characteristics of paper designed a semi-supervised learning algorithm to classify the candidates, the main advantage of the semi-supervised learning algorithm can take advantage of a small number of easily identifiable sample, the classification of a large number of difficult identification of samples. Finally, improved ID3 algorithm and semi-supervised learning algorithm based on to establish graduate swap system model, based on the establishment of the Graduate swap system model design and a simple swap. The core functionality of the system include: the conditions according to the candidates to recommend appropriate levels of schools, school information and inquiry. Test by test data, test results and the real results of the swap agreement, the model has the feasibility and use promotional value.