Association Rules Mining Based On Genetic Algorithm

In the last decades, the database technologies and magnanimity memory technologies have developed much. 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 arose at the historic moment and can be developed rapidly. Recently, data mining has been one of hot research area. The knowledge discovered by data mining technologies can be used to offer decision support.Association Rules Mining found a large amount of data associated with or related to the focus of the link between interesting. It is an important issue in data mining, then has been extensively studied by the filed in recent years. Rules research helps to find the link among different items in transactions database, find out customer buying behavior mode, the results of the analysis can be applied to merchandise shelf layout, survival mode, as well as to classify the users under the purchase mode. Agrawal was first proposed in 1993 equivalent to mining association rules about sets of items in customers, transaction database. After then many researchers made a lot of research on the issue of mining association rules. According to the requirement and the character of association rule mining, combine the idea of genetic algorithm, a mining method of association rules is proposed based on genetic algorithm. According to the analysis of example, it is a practical algorithm.The primary research of the paper was as follows:Firstly, this paper comments data mining, explains what is data mining, why and how to do it, clarifies the relation among data mining, machine learning, statistics, etc, the main process ofdata mining, classification and so on. Then, an important algorithm-----Genetic Algorithm isintroduced, and the origin, development, main theory, characteristic and extensive application of GA are briefly described.Deeply analyzing and studying association rule. In this paper, we have a systematic research into the measure standard for association rules. Studied the limitation in model based on support and confidence measure, the lift measure is given in this dissertation, which is used to prune the no-interest rules in order to discover the real interest rules mode.On this basis, this paper brings a forward mining method of association rules based on Genetic Algorithm, discusses and analyses the genetic algorithms in detail from coding method, fitness function, crossover operators, mutation operators and other aspects. Associated with the Marketing System, this paper gives the algorithms and program of mining association rule based on genetic algorithms.