基于免疫遗传算法的决策树分类分析-计算机软件与理论专业论文.docxVIP

PAGE PAGE VI Abstract Data mining has become a very popular research field, which is widely used in scientific research, business forecast and so on. It tries to extract information of potential, interesting and useful. Classification is one crucial branch of data mining. It’s a process of giving data object partition according to the characteristics of a group of data objects. This thesis aims to the classification techniques of decision tree. Decision tree learning using recursive top-down approach, it is similar to the growth of trees, through layers of branches, and it has formed a number of leaf nodes ultimately, each leaf node represents a decision-making categories. Genetic algorithm is an overall random searching method based on the theory of Darwinian natural selection and biological evolution. Immune genetic algorithm is a proved algorithm because it has overcome the premature convergence of Genetic algorithm, and improved overall search capability In this thesis, the immune genetic algorithm is applied to the generation of decision tree. It has achieved very good results. The main researches and contributions are as follows: This thesis has expounded the basic principle and method of decision tree at first, and sum up the advantages and disadvantages between two typical algorithms of decision tree. We expound the basic principle and method of Genetic algorithm and immune genetic algorithm, and sum up the advantages and disadvantages between them. Genetic algorithm has be applied to the decision tree classification method, the global search capability and parallel processing capability of genetic algorithm makes the decision tree algorithm converge become faster. Taking into account the evaluation of decision tree, the system of evaluation has proposed according to the accuracy, the support, the concise, and the information gain ratio. It has become as 基于免疫遗传算法的决策树分类研究fitness of genetic algorithm. Then the thesis has gave comparison of experiment, the e