L Company Demand Forecasting For Accessories Of Server

In recent years,China is speeding up efforts to formalization,leapfrog development of information industry has been realized.Key industries such as finance,telecommunication,traffic and education have been gradually rising with demands of server.L Company is a famous server manufacturer.In server supply chain which L Company become the center,one of the main sectors that impact company's yield is purchasing server accessories to upper vendor.This paper will research on L Company purchasing demand for accessories of server.By choosing the methods of classification and forecasting those are more suitable for L company,the classification of accessories will more meticulous,the forecast result will more accurate.L Company's procurement department can purchase server accessories on the basis of the forecast results after optimized.In this paper,on the basis of the relevant theoretical research,the current situation of L Company server accessories has been analyzed.The current situation analysis includes the current situation of classification and forecasting.On the accessories'classification issue,there are two problem,those are cross classification and low customer satisfaction.Because of inappropriate classification and unreasonable prediction method,accessories' forecasting results are in greater deviation from the actual values.For the existing problems of L Company server accessories demand forecasting,this paper mainly from two aspects of research:one is accessories'classification optimization,the other is accessories' demand forecasting optimization.On the basis of qualitative analysis,two phases model of demand forecasting optimization is presented in this paper.First phase of L Company server accessories is set up for classification.This method adds order date,purchasing time and times of order on basis of Pareto classification.In this part,ID3 algorithm has been used.Second phase set up ARIMA-BP neural network model.This model is based on traditional time series forecasting.It adds prediction of non-linear characteristic data,which makes the predict result closer to actual value.In order to optimize accessories' classification and forecasting results of purchasing demand,this article determined L Company server accessories forecasting results of demand.In this paper,data have gotten by comprehensive investigation.Firstly,using MATLAB software to solve server accessories classification problem in first phase and get more detailed classification of accessories.Then using SPSS software to solve ARIMA model,using MATLAB software to solve BP neutral network.Linear data plus no-linear data is forecasting results after optimization.According to the model,L company server accessories purchasing demand forecasting optimization are further analyzed.On the one hand is verify the model is practical;on the other hand,the results of accessories purchasing demand forecasting optimization do improved.The research indicates that the demand forecasting results after optimization is more closer to actual value than those results before optimization.