Research on the Prediction Model of Material Cost Based on Data Mining

Abstract

Material cost prediction should be based on the scientific mathematical models, to reduce the influence of subjective factors on the quota and other indicators of decomposition. This paper analyzes the particle swarm optimization (PSO) algorithm to optimize the parameters of support vector machine and establishes the prediction model of material cost after preprocessing the actual data and using the support vector regression (SVR) machine to perform data mining. In the forecasting process, the total cost of material is predicted, the predicted results are fitted with the actual value, and the relative errors are tested. The result shows that the forecasting effect is satisfied. The prediction of material cost is utilizing special methods to estimate and predict the level of material cost on the basis of historical data and relevant information. Its characteristic is predicting the future on the basis of history, and predicting unknown level on the basis of known information. The renewal of statistical data and a series of characteristics of material production of enterprises determine that the sequence formed by material cost data generally has the nonstationarity, nonlinearity and the point of abrupt change. The support vector machine can realize the minimization of structural risk, and the error rate on test data (namely generalization error rate) of learning machines takes the sum of training error rate and a dependent item as boundary. The support vector machine doesn't utilize the internal problems of the field, which can provide the good generalization performance in problems of pattern classification, and that is the specific characteristic of support vector machine. This paper conducts the data mining to establish the prediction model of material cost according to the learning method of support vector regression machine. This method is a type of relational schema between the spatial pattern of learning input and function mapping of learning output, and researchers generally call the function set in this type of mapping relation as learning machine. It starts from the research on observation data (namely sample). Researchers obtains some rules that can't be obtained from principle in the current situations, and meanwhile utilizes these rules to analyze the data obtained. Thus they can reach the value prediction and conduct the decision making and value estimation. 2. PSO-SVR MODELING ALGORITHM