Research On Early Warning Technology Based On Data Mining

The crisis has occurred inevitability and objectivity. Faced with the crisis, the best way is to eliminate the crisis nipped in the bud by a warning pre-control mechanism. Therefore, it is very important to adopt the appropriate theories and methods of early warning. Based on those theories and methods, some measures can be taken to deal with the crisis.Early warning of crisis is a kind of semi-structured or unstructured problem with a large number of uncertainties. It is difficult to scientifically assess and calculate because of a lot of factors without historical data and the corresponding statistics. There are many limitations in the traditional analysis methods, whose results of analysis and forecasting is not accurate enough.Data mining is a popular technique in the area of artificial intelligence and database; it can automatically analyze massive data in the data warehouse for mining rich and objective warning knowledge which can be applied in crisis warning. It has important theoretical significance and practical value to apply data mining techniques in the crisis early warning area.The main work of the dissertation is as follows:(1) The traditional early warning theory, methods and its applications are analyzed, and an early warning mechanism based on data mining techniques is proposed in the dissertation.(2) An early warning knowledge representation method based on domain knowledge constraints is designed. After analyzing the method of mining association rules, an early warning knowledge mining method based on domain knowledge constraints is proposed.(3) Utilizing one step strategy and iterative strategy, an early warning method based on early warning knowledge is designed to generate early warning information.(4) Aiming at the early warning problem of student's achievement, a useful research is explored by applying the early warning mechanism and method in a university educational administration management information system.