时间序列分析法在北京市食物中毒预测中的应用

Application of time series analysis in forecasting food poisoning in Beijing

目的探索时间序列分析方法在北京市食物中毒预测中的应用,为食物中毒的预防和控制提供依据。方法收集2004-2015年北京市食物中毒发生人数,采用ARIMA模型进行拟合,用2016年的事件数据验证模型拟合效果,并预测2017年北京市食物中毒发病人数。结果 ARIMA(1,0,0)×(1,1,0)4较好的拟合了既往时间段食物中毒发病人数的时间序列,拟合平均相对误差为6.00%,实际值均在预测值的95%CI内,预测2017年北京市食物中毒发生人数为264人。结论 ARIMA模型能够较好的拟合食物中毒发生趋势,在食物中毒发病人数预测中具有一定的应用价值。

Objective To explore the application of the time series analysis in the prediction of case numbers in food poisoning,and to provide scientific data for the development of food poisoning prevention and control strategies. Methods The ARIMA model was developed based on the seasonal data of food poisoning from the first quarter 2004 to the fourth quarter 2015,and then was tested by the data in 2016. Results Established ARIMA（ 1,0,0） ×（ 1,1,0）4 model was the optimal model with an average prediction fitting relative error of 6. 00%. The actual value was in the range of 95% confidence interval of the predictive value. The number of food poisoning in 2017 was predicted to be 264. Conclusion ARIMA model fitted well to trend of food poisoning and can be used for the short-term prediction in future.