The Study On Sale Forecasting Decision Support Systems (SFDSS) Based On Data Mining

Sales forecasting is the pivotal segment in supply chain management of enterprises, according to the result of sale forecasting, enterprises can just make rational materials procurement plan, production schedule, staffing plan, stock plan and marketing plan. So sale forecasting DSS used for supporting sale decision and other decision which are important for the development of enterprises. This thesis have been doing the research and exploration of the following aspects in the decision support of sales forecast based on data warehouse and data mining.(1) Research on sale forecast, DSS, data warehouse, data mining etc in great detail. Then introduce the technologies of data warehouse and data mining in the sales forecasting DSS, and design the sales forecasting DSS based on data warehouse and data mining.(2) According to the demand analysis and function definition of the system, design the overall architecture of the DSS and the function models. And propose the problems which can be solved by the technologies of data warehouse and data mining in the system.(3) According to the demand, propose the theme of sales forecast , carry on the design of the data warehouse , including conceptual design , logic design , physics design , data prepare etc. Finally to analyze the sale data by the way to cut slices of the data warehouse.(4) According to the goal of data mining, choose suitable data mining algorithms. Research on different ways which can solve the problems in traditional sales forecasting system, especially the ways that decision tree, neural network, related rule algorithms etc.(5) Carry on the instance analysis, build the data warehouse, use the method of neural network to analyze on selling historical data, then use the trained neural networkto predict sales amount and sales profit under the specified conditions. Verifysystematic validity.The system design is question-driven, intelligent, and has friendly human-computer interaction ability. The system can help administrator or analyzer to find the potential rule and issue among the data, and to predict the sales situation in the future. For example, it can predict sales amount, saleroom, price etc; it can analyze the customer's favorite, the hot-season of products, classify customers etc. The result of prediction can help to promote the sales of products, increase enterprise profit, and give decision support to enterprise for production schedule, purchase plan , stock plan , promote plan ,etc.The innovative point of this thesis is that it applies data warehouse, data mining technology to sale forecasting DSS, putting forward the ensemble structure and function module of the system, bringing forward the problems the data warehouse and data mining can solve in sale forecasting, and carrying on real example analyses, to show the process of constructing the system and prove the validity of the system, so as to reflect the value actual meaning of the study.