Ga-based Classification Rule Mining Technology Research And Application

Data mining technology is the integration of multidisciplinary technologies; these technologies include database and warehouse technology, statistics, machine learning, artificial intelligence, etc. The prime goal of data mining is to discover valuable information and knowledge which hidden in numerous data. Classification of Data Mining train samples from source data, which has class ID identifies. It summed up the relationship between identified class and non-identified class to identify unknown class data classification. Genetic algorithm is a global random search optimization algorithm.In this paper, I researched a classification of data mining using genetic algorithms. Firstly, I reviewed the data mining process; the data mining's basic concepts and basic theory, emphasizing to discuss the Concepts, processes, basic technology and its evaluation criteria. Then, discussed genetic algorithms, analyzed the genetic algorithm and the relationship between natural selection, the basic method of simple genetic algorithm, genetic algorithm theory and existent problems.Through the research of data mining and genetic algorithm, I found the common problem in data mining when use genetic algorithm. For example, the classification rules are conflict, the efficiency of the large database, the premature convergence of genetic algorithms and genetic encoding issues. In order to resolve those problems, I used a conflict resolution strategy to solve the conflict between the rules. I used 3-exchange heuristic crossover operator and the 3-exchange mutation method to solve the problem of premature convergence. A genetic algorithm based on matrix decoding algorithm for classification rules is proposed for low efficiency of the larger database. Finally, I use the car evaluation data to verify it. The algorithm and the J4.8 algorithm are compared by the four evaluation criteria of the classification algorithm, the results shows that the algorithm is superior to J4.8 algorithm; It Proved the feasibility and effectiveness of this algorithm.