SARIMA等模型在乌鲁木齐市其他感染性腹泻病发病预测中的应用

Application of SARIMA and other models in the incidence prediction of other infectious diarrhea diseases in Urumqi

目的构建乌鲁木齐市其他感染性腹泻病的时间序列模型,比较季节性差分自回归滑动平均(seasonal autoregressive integrated moving average,SARIMA)模型、霍尔特温特斯(Holt-Winters)模型和先知(Prophet)模型的预测能力,为乌鲁木齐市其他感染性腹泻病的防控提供理论参考。方法根据2010年1月至2021年12月乌鲁木齐市其他感染性腹泻病的发病率分别建立SARIMA模型、Holt-Winters模型和Prophet模型,比较三种模型的拟合效果,以均方根误差(root mean squared error,RMSE)、平均绝对误差(mean absolute error,MAE)、平均绝对百分比误差(mean absolute percentage error,MAPE)、决定系数(coefficient of determination,R^(2))作为评价指标选取最优模型。结果季节指数显示,5~10月为乌鲁木齐市其他感染性腹泻病的流行季节;SARIMA(0,0,3)(0,1,1)12模型拟合RMSE为2.54、MAE为1.56、MAPE为0.28、R^(2)为0.71;Holt-Winters模型拟合RMSE为3.32、MAE为2.06、MAPE为0.33、R^(2)为0.54;Prophet模型拟合RMSE为3.37、MAE为2.15、MAPE为0.44、R^(2)为0.48;SARIMA(0,0,3)(0,1,1)12模型预测RMSE为4.26、MAE为3.64、MAPE为0.46。结论三种模型中SARIMA(0,0,3)(0,1,1)12模型拟合效果最好,能较好的捕捉乌鲁木齐市其他感染性腹泻病发病趋势,对乌鲁木齐市其他感染性腹泻病科学防控具有一定指导价值。

Objective To construct a time series model of other infectious diarrhea in Urumqi City and compare the prediction ability of the seasonal autoregressive integrated moving average(SARIMA)model,Holt-Winters model,and Prophet model to provide theoretical references for the prevention and control of other infectious diarrhea in Urumqi City.Methods The SARIMA model,Holt-Winters model,and Prophet model were established based on the incidence rate of other infectious diarrhea diseases in Urumqi from January 2010 to December 2021,and the fitting effects of the three models were compared.The root mean square error(RMSE),mean absolute error(MAE),mean absolute percentage error(MAPE),and determination coefficient(R^(2))were used as evaluation indexes to select the optimal model.Results The seasonal index showed that from May to October was the epidemic season of other infectious diarrhea diseases in Urumqi.SARIMA(0,0,3)(0,1,1)12 model fitting index RMSE was 2.54,MAE was 1.56,MAPE was 0.28,and R^(2) was 0.71.The RMSE,MAE,MAPE,and R^(2) fitted by the Holt-Winters model were 3.32,2.06,0.33,and 0.54,respectively.The Prophet model fitting index RMSE was 3.37,MAE was 2.15,MAPE was 0.44,and R^(2) was 0.48.SARIMA(0,0,3)(0,1,1)12 model predicted RMSE was 4.26,MAE was 3.64,and MAPE was 0.46.Conclusion Among the three models,SARIMA(0,0,3)(0,1,1)12 model has the best fitting effect,which can better capture the incidence trend of other infectious diarrhea diseases in Urumqi,and has certain guiding value for the scientific prevention and control of other infectious diarrhea diseases in Urumqi.