Study On The Improved C4.5Decison Tree Algorithm And Its Application On The Student Performances Prediction

In recent years,data mining has caused the information industry and the great attention of the society,the main reason is the presence of large amounts of data can be widely used,and is in urgent need of these data into useful information and knowledge. Access to information and knowledge can be widely used in various applications,such as market analysis,customer retention,scientific research,etc.In this process,the data classification is an important topic on data mining. Currently there are numbers of data classification methods; the decision tree method,due to its clarity on theory,easily to understand and transfer to classification rules,is widely be studied and applied.In this article,we present a "student performance management" system; we studied how to apply the data mining technology on our recently available database system,abstract the useful information that hiding among the huge data,and provide the school managers a complete analysis. Based on our studied on data mining,we developed a module for student performance prediction; in this module,the with improved C4.5algorithm,we developed a decision tree on the student performance database,and predicted the student scores for the college entrance exam.By analyzing several typical decision tree algorithm analysis and comparison,in this article puts forward an improved C4.5algorithm. The algorithm is of Higher Mathematics in the some principle and C4.5algorithm combining on the algorithm,the information entropy and split information quantity formulation for simplification,in order to improve the algorithm running efficiency.We use the program implemented both the original C4.5algorithm and the improved C4.5algorithm,and then comparative experiments. Base on the experiment result,we concluded that the time needed for building a decision tree is less for the improved C4.5algorithm. The improved C4.5algorithm in this article comes with better performance and shows impressive results on classification.