Research And Application Of MUlti-objective Optimization Design In Data Mining

With the extensive application of database, the amount of data that people has accumulated is increased rapidly, and it is the urgent affair how to extract useful knowledge from a mass of data. In this case, the idea of data mining is introduced, it can predict future trends and behaviors, and can make advanced knowledge-based decisions.At present data mining technology has been widely applied in every field, for example, the results of data mining can optimize the decision-making. However many optimization decision-making problems become more complicated in real application, so there are important academic significance and application value to solving multi-objective optimization problem(MOP) through effective methods.Basing on data mining technology and multi-objective optimization design theory, this paper mainly studies MOP with a concrete example. First, the background of selected topic and significance of study are introduced, the domestic and foreign research situation of data mining and multi-objective optimization design are analyzed, meanwhile the main research contents and structures of the paper are given. Second, data mining technology is introduced simply, and the basic concepts of data mining are given, the common algorithms of data mining are presented, and the applications of data mining are enumerated. Then, the emergence and development of optimization design are described briefly, the definition and the principle of multi-objective optimization design are given, meanwhile the characteristics of Genetic Algorithm (GA) are analyzed, several methods based on GA for solving MOP are introduced emphatically, and some classical cases are enumerated to illustrate the uses of multi-objective optimization design. And, a Pareto GA method to deal with MOP is presented for an Eomecon Chinanthe Alkaloids (ECA) against Oncomelania, and the fitness function is built, then the ranked-select method is applied to transform constraint MOP into non-constraint MOP with an immigration operator which improves the computing performance observably, a Pareto optimal set can been got, the result proves the efficiency and advantage of this method. Finally, the main research contents are summarized, and the further prospect of research is predicted.