时间序列分解法在我国食物中毒发病人数预测中的应用

发布时间：2018-06-06 09:04

本文选题：ARIMA模型 + 分解分析法　；参考：《中国卫生统计》2015年04期

【摘要】：目的对ARIMA模型和时间序列分解预测方法在我国食物中毒发病人数预测中的效果进行比较,探讨优化模型,为更好地了解我国食物中毒发病人数提供预警和参考依据。方法收集2000-2013年我国食物中毒季度发病人数,用Excel 2003和SPSS 20.0拟合ARIMA模型和时间序列分解预测模型,用2013年的数据评价模型的预测效果,并对2014年各季度食物中毒发病人数进行预测。结果两种方法预测食物中毒发病人数的R2分别是0.355和0.919;MRD分别为34.350%和14.507%;MER分别为0.303和0.110;MSE分别为293505.000和43570.000;RMSE分别为541.761和208.736;MAE分别为413.500和149.500;预测的2014年各季度食物中毒发病人数依次为387、1020、1357、606。结论时间序列分解法预测效果优于ARIMA模型,可以用来预测我国食物中毒的发病人数,预测效果可靠。
[Abstract]:Objective to compare the effects of ARIMA model and time series decomposition prediction method in predicting the number of food poisoning cases in China, and to explore the optimized model to provide early warning and reference basis for better understanding of the number of food poisoning cases in China. Methods the quarterly incidence of food poisoning in China was collected from 2000 to 2013. ARIMA model and time series decomposition prediction model were fitted with Excel 2003 and SPSS 20.0, and the prediction effect of the model was evaluated with the data of 2013. The incidence of food poisoning in every quarter of 2014 was predicted. Results the R2 of the two methods for predicting the incidence of food poisoning was 0.355 and 0.919 MRD were 34.350% and 14.507%, respectively, 0.303 and 0.110 MSE of 293505.000 and 43570.000 were 541.761 and 208.736%, 413.500 and 149.500, respectively, and the predicted incidence of food poisoning in different quarters of 2014 was 3871020 / 1357606, respectively. Conclusion the time series decomposition method is superior to the ARIMA model and can be used to predict the incidence of food poisoning in China.
【作者单位】：华北理工大学公共卫生学院;
【分类号】：R155.3

【参考文献】