The Study On Food & Beverage Sales Forecasting System Based On Data Warehouse And Its Application

With the foundation of Food&beverage Enterprise Information System, a large amount of data related to their business operation system in different enterprise'departments is accumulated and it is lack of organization and utilization. In result, Food & beverage business decision-makers need to figure out the hidden information including in the original data in order to support their decision system. At the same time data warehouse technology caught more and more attention leads a new way to create decision support system .Basing on the fully understanding the demand of users, this article proposes to create a kind of sales forecasting system based on the data warehouse technology, its prime task is as follows:Firstly, data warehouse model research: based on analysis to the Food&beverage enterprise's demand, proposes the data warehouse three stage standardization model description integrated model method: namely conceptual model design, logical model design physical model design three stages; In order to guarantee the convince of the data conversion and the data loading, proposes to a method based on the meta data principal linkage mapping proxy to clean process to solve the interior isomerism data problem.Secondly, basing on time series decision tree classification migration average forecasting model establishment: the forecasting data in the Food&beverage enterprise has some profession characteristic, such as frequent big quantity and seasonal characteristic, so designs a tree classification migration average forecasting model base on time series . The essence of the model is to classify the historical sale data of the Food&beverage enterprise related to the decision tree technology and the sale characteristic in the Food&beverage enterprise and to form smog rules of classification, so the forecasting time spot can to classify the corresponding kind, Thus in the same characteristic sales data class carries on the motion average to obtain the forecasting data automatically. Comparing with other complex time series forecasting methods, the method has??e profession characteristic ,such as simple, conforming to the profession characteristic, the strongly operation practicality and so on,at the same time it eliminates the influences because of undulatory property and seasonal characteristic.Thirdly, through to the forecasting application in a Food&beverage enterprise, according to comparing the forecasting sales data with the actual sales data and the sales data which obtains from migration average forecasting model , the progress of comparing has confirmed the serviceability, the validity and the sophistication of decision tree classification migration average forecasting method based on the time series.A prospective study was done in this study to enhance the efficiency of decision-making system in the field of Food&beverage enterprise information and its ordinary application , and in this paper it proposes to establish a decision tree classification migration average forecasting method based on the time series to mine the historical sale data to be used for the prediction of sale achievement in the future, at the same time the method can be applied in other field because of its serviceability and originality.