摘要
卡尔曼滤波算法是一种最优线性递推估计方法,受数据分布特点影响小,适应范围广,建模简单,适合于对各种复杂时间序列的预测,效果显著。鉴于四川省社会消费品零售额数据分布不光滑,运用卡尔曼滤波算法对之进行了预测,取得了很好的效果,平均预测误差仅0.772406%,比ARMA模型的平均预测误差2.1323%减小了63.7756%。由模型预测得到2019年四川省社会消费品零售额为21570.26亿元。
Kalman filtering algorithm is an optimal linear recursive estimation method,is less affected by the characteristics of data distribution,and has a wide range of applications and simple modeling.It is suitable for prediction of various complex time series with remarkable effect.In view of the uneven distribution of the retail sales data of social consumer goods in Sichuan Province,Kalman filtering algorithm was used to predict the retail sales data and achieved good results.The average prediction error was only 0.772406%,which was 63.7756%lower than the average prediction error of ARMA model of 2.1323%.The model predicts that the retail sales of social consumer goods in Sichuan Province in 2019 will be 21570.26 billion yuan.
出处
《阿坝师范学院学报》
2019年第4期43-48,共6页
Journal of Aba Teachers University
基金
湖北省自然科学基金项目“信息挖掘与信息融合技术研究”(2017CFB164)
关键词
四川
社会消费品零售额
预测
卡尔曼滤波
ARMA
Sichuan
retail sales of social consumer goods
forecast
Kalman filtering
ARMA
