The Establishment And Application Of Support Vector Machine Forecasting Model

Support vector machine (SVM) is a kind of novel machine learning method based on statistical learning theory (SLT). It is assured that SVM has many advantages, such as the global optimization, the most generalization ability, and the strong extension etc in terms of structural risk minimization (SRM). SVM can solve many practical forecast problems. It has become one of the most influent achievements in machine learning area.This paper primarily constructs several forecasting models about SVM:(1) Grey support vector machine prediction model (GSVM). Grey support vector machine (GSVM) model is established by grey cumulative theory and support vector machine fitting nonlinear data capabilities. The forecasting values are obtained by reduction. The result shows that the model has higher prediction accuracy.(2)The autoregressive model, BP neural network model and support vector machine model are studied in the paper, respectively. Taking our country population growth rate as a study case, the prediction results are obtained by three methods. Analyzing the characteristic of combining forecasting method, the Theil coefficient optimization combination forecasting model based on neural network and SVM forecasting model is set up. The optimization combining model is applied to predict population growth rate.(3) Owing to SVM can catch the data nonlinear characteristic. SVM combining forecasting model based on autoregressive model, BP neural network model, and support vector machine model is built. The combining model is applied to predict population growth rate. Compared with single prediction model and the Theil coefficient combination forecasting model, the results show the precision of the SVM combination forecasting method is the highest, which can solve the nonlinear problem of predicting population growth better.