An Improvement Of ID3 Alogrithm Based On Genetic Algorithm In Data Mining

Along with the modern information technology rapid development, many domains all accumulated the massive data. Longed for discovered latent knowledge and rule in these data accomplishes the data mining discipline to start and the data mining technology development. As a multi-disciplinary overlapping comprehensive domain, the data mining has involved the database, statistics, the machine learning, the high performance computation, the pattern recognition, the nural network and the data visible and so on.The decision tree algorithm is a method induction studying a group of known demonstrations, and generates a decision tree. The most typical decision tree sorter study algorithm is the ID3 algorithm, it uses the strategy which divides and rules from the top, uses the information gain the standard choice fission attribute, can guarantee the structure leaves a simple tree. This algorithm is simple and highly effective, the production knowledge was easy understood by the person, but has the question of excessively fits when faces the massive data to gain the knowledge. Proposed a ID3 algorithm based on the genetic algrithm in the foundation of thoroughly analyzing the ID3 algorithm and build a mode on the data which has the invasion data. First of all, this algorithm recognizes the rules as gene, carries on the evolution, divides the rules set using genetic algorithm, and then generate the decision trees group using the rule set, at last gives the result using the decision trees group. The experimental result had indicated, this algorithm could better classify, and compared with the result produced by the ID3 algorithm.