Study On Forecast Arithmetic In Decision Support System

This paper places great emphasis on the study of the forecast arithmetic, computing intelligence and its application in "Decision Support System (DSS) on economic and social development of cities". DSS is a people-computer system, it based on manage, operational research, cybernetics and behavior research;its instruments are computer, emulation and information technology;aims at half-structured question;bears decision-making and has aptitude function. At present, many fields are composing their DSS hardly, and the DSS has a big development. However, if we want a correct decision, its premise is correct forecast. So the forecast method, method implement and the precision of forecast are core of DSS. The research focus on meliorating forecast arithmetic, improving its precision and realizing the forecast system.In this paper, through analyzing computer progress of regression and time serials firstly, put forward means for advancing precision in practical process, which including choosing proper samples, functions, and parameters. Validate these methods using real data, verifying precision and validity of these methods.secondly, through combining traditional forecast arithmetic with computing intelligence, introduces Artificial Neural Networks to solve the problem of Time Series Prediction. Make a further research on Time Series Prediction using BP 1 neural network, including general steps of building a model, processing samples and selecting best parameters in reason. Then build a forecast mode based on real dates, and final results prove this neural network can advance the precision on some degree.At last, program these methods, finally design and realize the forecast system. The system includes four parts: dealing dates, managing forecast methods, building models, and forecasting results. Implement this system into "DSS on economic and social development of cities", we can forecast all indexes in it.