生猪市场价格风险预警的GRNN建模及其实证研究

为了对生猪市场价格风险进行预警,根据我国2009年1月-2011年8月14个指标的32组样本数据,建立了广义回归神经网络(GRNN)预警模型,其中训练样本29组,测试样本3组.训练样本和测试样本的均方根误差、平均绝对误差(AAE)和相关系数都非常接近,说明建立的模型具有较强的泛化能力和鲁棒性,测试样本的AAE为0.0062,平均相对误差为2.3%,说明建立的GRNN模型具有很高的预测精度,可用于我国生猪市场价格风险预警研究和实际预测,并为政府有关部门指导生猪生产和进行市场调控提供决策依据.

Forecasting the Market Price Risks of Live Pigs Applying a General Regression Neural Network and Its Empirical Research

In order to forecast the market price risks of live pigs,the general regression neural network (GRNN) model was applied and established according to the thirty-two sample data of fourteen indexes during 2009.1-2011.8.Of the sample data,twenty-nine is training data set (TrDS),and the other is testing data set (TeDS).The values of the root-mean-square error (RMSE),the absolute average error (AAE) and the relative coefficient R between the TrDS and the TeDS are very close,and show that the GRNN's generalization and robustness is quite good.The forecasted AAE of the TeDS-the market price risks of live pigs in three months,is less than 0.0062,and its mean absolute percentage error is less than 2.3％.These results show that the above-established GRNN model possesses good practicality,and it can be applied to forecast the market price risks of live pigs,administrate the live pigs' production,and adjust the market by the local government and its authorities.