论文部分内容阅读
目的采用季节时间序列模型(SARIMA)对福建省细菌性痢疾发病数进行短期预测,为风险评估提供可靠的定量数据基础。方法运用EViews 8.0软件,基于SARIMA模型,对福建省2004年1月至2015年12月细菌性痢疾月发病数进行分析和建模,再对2016年1~9月进行预测和评估,然后修正模型对2016年10~12月进行短期预测。结果 2004年1月至2015年12月,福建省细菌性痢疾月发病序列呈下降态势和周期性波动。SARIMA(0,1,1)(1,1,1)12拟合优度较好,预测准确度和精度较高,均方根误差(RMSE)为26.59,平均绝对百分比误差(MAPE)为13.61%。2016年1~9月前瞻性长期预测值MAPE为19.44%,其中7~9月MAPE为20.49%,而2016年7~9月前瞻性短期预测值MAPE为6.48%,而且标准误(SE)小于长期预测。采用2004年1月至2016年9月细菌性痢疾例数进行建模拟合后,SARIMA(1,1,2)(0,1,1)12为最佳模型,2016年10~12月短期预测结果分别为41例、36例和24例。结论 SARIMA模型能够对福建省细菌性痢疾发病数进行较准确的短期预测,可为风险评估提供可靠的定量数据基础。
Objective To estimate the incidence of bacillary dysentery in Fujian Province using the seasonal time series model (SARIMA) and provide a reliable quantitative data foundation for risk assessment. Methods EViews 8.0 software was used to analyze and model the monthly incidence of bacillary dysentery in Fujian province from January 2004 to December 2015 based on SARIMA model. Then, the forecast and assessment of January, September, Short-term forecast for October-December 2016 Results From January 2004 to December 2015, the monthly incidence of bacterial dysentery in Fujian Province showed a decreasing trend and cyclical fluctuations. The goodness of fit of SARIMA (0,1,1) (1,1,1) 12 is better, and the prediction accuracy and precision are higher. The root mean square error (RMSE) is 26.59 and the average absolute percentage error (MAPE) is 13.61 %. From January to September 2016, the prospective long-term predictive value (MAPE) was 19.44%, of which MAPEN was 20.49% from July to September 2016 and from July to September 2016, the prospective short-term forecast MAPE was 6.48% and the standard error (SE) Long-term forecast. SARIMA (1,1,2) (0,1,1) 12 is the best model after modeling the number of cases of bacillary dysentery between January 2004 and September 2016. In the short term from October to December 2016 The predicted results were 41 cases, 36 cases and 24 cases. Conclusion The SARIMA model can make a more accurate short-term prediction of the number of bacillary dysentery in Fujian Province and provide a reliable quantitative data foundation for risk assessment.