Research On Application Of Genetic Algorithem In Data Mining

Recently, the database technologies and magnanimity memory technologies have developed much. The ability of people to collect data has been further improved. At information age, oriented to a great deal of data, how to utilize the huge original data to analysis the current situation and predict the future effectively, have already become a great challenge that the mankind has faced. Therefore the data mining technology is increased at the historic moment and can be developed rapidly. Nowadays, data mining has been one of hot research area. The main works in this paper as follows:First, the basic definition and its principles of data mining and genetic algorism are introduced. Then the typical data mining algorithms and the classical genetic algorithms are analyzed, include its advantages and disadvantages. The basic definition and basic principle of is proposed, which supplies the theoretical basis for continuous research.Second, we present a method of improve genetic algorithm on clustering in data mining (IGAOC) .Based on the supermarket's marketing system, according to the supermarket's sales for a period, the database information is analyzed on the basis of improved clustering of genetic algorithm, the types of goods selling better is predicted, the analysis some potential applied value is analyzed, and a certain guiding point is provided for the supermarket's marketing development. The experiment shows that this algorithm owns better quality and comprehensive ability than k-means algorithm to analyze clustering.Third, a method of data mining based on immune genetic algorithm is proposed. It is based on the using applied genetic algorithm on the date mining. Carrying out improvement on the Genetic algorithm and introducing immunity operator (IGA). We can use this algorithm to achieve Customer Relationship Management (CRM). Take the result mined by that above-mentioned IGAOC method as example. For the same kind of commodities, there is more than one supplier, so this method can mined the basic of all supplier information, and calculate the best supplier who will make the supermarket to a maximal profit. Then lock these suppliers and take maximal benefit for the supermarket. The experiment shows that the algorithm has a strong robust and implicit Parallelism. It can be used for rapid, effective and thorough search. And it is an effective method to deal with large-scale data. It not only overcomes the phenomenon of premature convergence in GA, but also improves the efficiency of searching greatly.