The Theory And Method Of Self-organizing Data Mining And Complexity Research Of Economy System

The complexity research of economy system emphasizes on revealing the cause of economy structure and its evolutive process, and forecasting the development of economy as accurately as possible. As long as we use scientific methods to open out the evolutive law of our complex economy, adopt feasible countermeasures, institute scientific middle and long run layouts, we can avoid fatal misplay in decision-making. Therefore, the complexity research of economy system is very important to the healthy development of our national economy.In this paper, we have studied and developed the theory and method of Self-organizing Data Mining thoroughly, and applied it to the complexity research of Chinese economy system. The major achievements in this paper are as follows:(1) A new viewpoint is presented: the method of Self-organizing Data Mining is a valid technique for realizing the method of the qualitative to quantitative meta-synthesis using to complex system. The demonstration research on the influence factors of RMB exchange rate has confirmed this viewpoint.(2) Self-organizing combination forecasting method is put forward. This method can improve the forecasting ability of model remarkably when comparing to other familiar combination forecasting methods. The example has shown its validity and maneuverability in forecasting.(3) Linking to the character of complex economy, the improvements of themethod of Self-organizing Data Mining are advanced. By adding data samples of same type, we have solved one contradiction in application of Self-organizing Data Mining algorithm: its Analog Complexing algorithm needs large data samples while there are only narrow data samples in complex economy system. Whereas we may pay more attention to the latest behavior of economy system, we have made corresponding mend to the minimum bias criterion in Object System Analysis. Enlightened by the move average index in technique analysis of security, when forecasting stock market, we have used the Self-organizing Data Mining algorithm with move average variables. The demonstrations have shown that the results gained from above improvements accord more with the practice of economy system.The result of this paper is the development of the theory and method of Self-organizing Data Mining as well as the complexity research of economy system, it has reference significance to the right understanding of evolutive law of our national economy system and to the scientific decision-making.