3种模型预测猪瘟发病率的比较

如何控制和消灭猪瘟一直是防控工作的难点,对猪瘟发病率进行早期预测是目前研究的重点方向.本研究拟用ARMA模型、指数平滑模型和季节指数模型对猪瘟发病率进行预测,采用MSE、MAE、MPE和MAPE来评价模型的精确度.结果显示,ARMA模型的4种评价指标均为最小值,预测值最接近真实值,对猪瘟发病率的预测效果较好.应用定量预测法对猪瘟发病率进行数值上的模拟,可较好的反映近几年猪瘟的流行趋势和发病率的变化,为制定最优防控策略提供理论依据.

The comparison of ARMA, exponential smoothing and seasonal index model for predicting incidence of classical swine fever

How to control or eradicate classical swine fever has always been a challenge. The early estimation to incidence of classical swine fever is the key research direction at present. In this study, ARMA,exponential smoothing and seasonal index model is adopted for predicting incidence of classical swine fever, and evaluating the precision of these three models using MSE,MAE,MPE and MAPE. The results showed that the four kinds of evaluation indexes of ARMA model are minimum, the predictive value of ARMA model is close to the real values, and ARMA model has the fittest predicting effect. The method of quantitative predictions is used in numerically simulating the incidence of classical swine fever that can reflect the transmission tendency and incidence's change of classical swine fever in recent years and provide theoretical basis for making optimum preventive strategies.