ARIMA模型和Holt-Winters指数平滑法在自杀死亡中的预测效果比较

Comparison of effects of ARIMA model and Holt-Winters exponential smoothing method in the prediction of suicide death

目的探讨比较自回归求和滑动平均(autoregressive integrated moving average,ARIMA)模型和Holt-Winters指数平滑法在自杀死亡预测中的应用。方法利用河北省2014年1月-2018年6月自杀月度死亡资料分别建立ARIMA模型和Holt-Winters指数平滑模型,对2018年7-12月自杀月度死亡例数进行预测,并与实际死亡人数进行验证比较,然后根据2个模型的均方根误差(root mean square error,RMSE)、平均绝对误差(mean absolute error,MAE)以及平均绝对百分比误差(mean absolute percentage error,MAPE)评价模型的预测效果。结果2014-2018年河北省累计报告自杀死亡人数2882例,自杀死亡水平整体呈现下降趋势,构建的ARIMA最佳模型是ARIMA(0,1,1)(1,1,0)_(12),预测结果的RMSE、MAE和MAPE分别为5.99、4.67和9.80%;Holt-Winters指数平滑法最佳拟合模型是乘法模型,预测结果的RMSE、MAE和MAPE分别为6.03、5.17和11.44%。结论ARIMA模型预测效果优于Holt-Winters指数平滑法,更适用于自杀死亡趋势的短期预测。

Objective To explore and compare the application of autoregressive integrated moving average(ARIMA)model and Holt-Winters exponential smoothing method in the prediction of suicide death.Methods The ARIMA model and the Holt-Winters exponential smoothing model were established respectively based on the data about the monthly death cases of suicide from January 2014 to June 2018 in Hebei Province.The monthly death cases of suicide from July to December 2018 were predicted and then compared with the actual number of deaths.And the prediction effects of the two models were evaluated based on root mean square error(RMSE),mean absolute error(MAE)and mean absolute percentage error(MAPE).Results A total of 2,882 suicide deaths were reported in Hebei Province from 2014 to 2018,and the overall level of suicide deaths showed a downward trend.The best ARIMA model established was ARIMA(0,1,1)(1,1,0)_(12),and RMSE,MAE and MAPE of the predicted results were 5.99,4.67 and 9.80%,respectively.The best fitting model of Holt-Winters exponential smoothing method was the multiplicative model,and RMSE,MAE and MAPE of the predicted results were 6.03,5.17 and 11.44%,respectively.Conclusion The prediction effect of ARIMA model is superior to that of Holt-Winters exponential smoothing method,and it is more suitable for short-term prediction of trend of suicide deaths.