The Study And Application Of Data Mining Method On The Optimization Of Procurement Process

The optimization of procurement includes the prediction on the raw materials in the production of the enterprise and subsequent decision making of procurement. The aim of the procurement optimization is to reduce the occupied capital for procurement and the cost of products. In modern enterprise management, the procurement capital takes great part of the turnover capital, so it is of great importance to make reasonable purchasing decisions.First the consuming amount of raw materials must be predicted. To be correct, the rule must be analyzed from large volume of data. From 1990s, data mining technology began to be applied to solve the business problem. In these years, due to the ceaseless advancement in computers' computing ability, the research direction of data mining starts to emphasize particularly on how to analyze data and get needed information by optimal searching algorithms. The work of this paper is how to optimize the procurement by the combination of data mining method and optimal searching algorithms.The problem is divided into two sub problems. One is the prediction of the consuming of raw materials, and the other is make optimal purchasing plan. For the first one, the consuming of the raw materials is considered as a probability model swayed by products. From the historic data of products' producing records, the parameters of the assistant materials' probability model could be worked out by the method of maximum likelihood. The problem of the solution-finding for the parameters is changed to a problem of constraint optimization. In this case the improved "complex method" algorithm is applied considering that it fits kinds of probability models. For the purchasing plan-making problem, on the basis of data analysis, the influencing elements are found and a procurement cost model is constructed. Then through the optimization for the model, the purchasing plan is made.The method referred in this paper can be used online, and independent of the statistics model, so that it is extendable and universal. By applying in the printing and dying enterprise for nearly half years, this method is proved to be helpful to deduce the rule of the consuming of raw materials from historic data, and the purchasing plan can be made, so that the purpose is realized.