参数优化支持向量机的人参价格预测模型

摘要

Through the analysis of the influence of ginseng price factors,we predict the ginseng price. We applied K-fold cross-validation method,utilized the PSO (Particle Swarm Optimization) to reach the optimum of penalty parameter c and value of ggamma,and built the forecast model of the ginseng price from January 2010 to December 2011. The optimized value by PSO for penalty parameter c is 3. 6974,the prediction correlation coefficient for prediction set 1 by the SVM (Support Vector Machine) of radial basis kernel function is 97. 316% ,the results are satisfactory, and we can predict ginseng price from January to June 2012 with this model.%为了对人参价格进行预测,分析了影响人参价格因素,通过K-fold交叉验证方法,利用粒子群算法对支持向量机的惩罚参数c和ggamma值进行寻优,建立起2010年1月～2011年12月林下参的价格预测模型.利用粒子群算法优化惩罚参数c为3.6974,利用radial basis function核函数的SVM(Support Vector Machine)对预测集1的预测相关系数为97.316％.