Research And Application Of Data Mining Technology On Macroeconomic Intelligent Decision Support System

As an emerging discipline, Data Mining makes the application of data from simple low-level queries, to mining the knowledge from data so as to provide decision-making support for people. Through Data Mining it can automatically process the mass data in data storage, abstract meaningful patterns, and find out the goal knowledge people need.Macroeconomic is an enormous system which constitutes of many interrelated economic elements which exist certain of causality. In a transition period for Chinese macroeconomic, due to the establishment and improvement of the economic system, the market norms and so on, the economic forecasting and decision-making is being more complex and meaningful.Relying on the province's major scientific and technological innovation project "Intelligent-decision supporting system based on macroeconomic data warehouse" which I participated in, this paper systematically expounded the relevant theory of Data Mining, and introduced four common used methods: the classification, regression, clustering and association rules. According to Data Mining's benefits and macroeconomic characteristics, the paper proposed Data Mining to be a macroeconomic forecasting method, and had designed and completed the Data Mining subsystem, which applied Support Vector Machine, Neural Network, Fuzzy C-means Clustering, the Least-squares and Apriori algorithm to macroeconomic research for Data Mining. Practical application shows that Data Mining can make good effects for economic forecast.Innovation points:1,According to Data Mining's advantages and macroeconomic characteristics of high-dimensional data, small samples, timing, multi-thematic and multi-level, it proposed Data Mining as a new method of macroeconomic forecasts;2,It applied Data Mining in macroeconomic analysis, extracted scattered economic data and conversed the data into a subject-specific data warehouse. It also used Data Mining methods of classification, clustering, regression and association rules to deal with the macroeconomic actual problem, organically combined economics and intelligent computing.3,A new Support Vector Machine fast learning algorithm based on border vector was proposed. This algorithm rate has a greater increase than the traditional Support Vector Machine method and demands for the calculation memory space are also significantly reduced. At the same time, it will not affect the performance of Support Vector Machine, because it doesn't lose support vector in the selection process for vector borders.