The Information Processing Method In The Analysis Of Student Achievement

With the development of nowadays science and education, a numbers of universities have enrolled tens of thousands students. In this context, college course scores have complex inherent relationships. With a collection of scores data, teachers should not only study these data with simple statistical operation, but also investigate the potential information initiatively with proper techniques, so as to transform data to available knowledge, which permits a better understanding of students behaviors and an improvement of teaching and learning qualities. This is a project that several universities have been taking into consideration. However, traditional analysis involves no more than average, variance computation and significance, confidence testing, which does not provide an efficient feedback with respect to learning aspects. Fortunately, advance in modern information processing gives birth to techniques such as clustering method, decision tree, canonical correlation analysis and support vector machine, etc. These novel methods can overcome shortcomings of traditional methods and can be used for extracting information from large scale data. They have found their applications in many domains and provided useful assistance for decision. Although researchers have attempted to use these methods in student scores analysis, there is still large room to explore. The study of proper ways to apply these methods in student score processing would be very helpful to understand relations embedded in score data and corresponding factors, which allows us to act according to the situation, so as to enhance positive factors and eliminate negative factors. Teachers may then improve teaching strategies and students could be guided by right learning strategies.This thesis explored student score analysis based on the advance of information processing techniques. At the beginning we summarized the state-of-art of score analysis methods and gave out the systematical processing procedures. After that, two methods were proposed respectively with regards to processing and analysis procedures. Concretely speaking, in the aspect of score processing, we considered a frequently used group grading strategy, which may result in the disagreement of score distributions and incomparability of scores from different groups. Firstly we discussed four strategies of rational group grading in principle. Secondly, based on information processing techniques we proposed a histogram specification method and gave out the operational steps. In the aspect of score analysis, we studied the canonical correlation analysis which aimed to investigate the correlation between two sets of variables. Its principle was introduced and potential applications in educational research were explored. The operational steps were also given for score analysis. After that, the nonlinear extension of this method was discussed in brief.Moreover, a number of college student scores were collected as the raw data. The author implemented the methods discussed in this thesis with the famous technical computing tool-Matlab, to demonstrate the steps of processing and the results of analysis. The proposed methods were validated by these experiments and useful feedbacks were extracted for improving teaching quality.