两种时间序列分析法在预防接种门诊接种量预测中的应用  

Application of two methods of time series analysis in the prediction of vaccination quantity in outpatient department

作　　者：常瑾[1] 葛彩云[1] 朱卫红 李立春 CHANG Jin;GE Cai-yun;ZHU Wei-hong;LI Li-chun(Preventive and Healthcare Station,Daxing District Xihongmen Hospital,Beijing 100162,China)

机构地区：[1]北京市大兴区西红门医院预防保健站,北京100162

出　　处：《中国公共卫生管理》2022年第1期64-67,共4页Chinese Journal of Public Health Management

摘　　要：目的比较指数平滑法和自回归滑动平均(Autoregressive Integrated Moving Average,ARIMA)模型预测预防接种门诊量的效果。方法收集2012—2019年北京市西红门医院预防接种门诊月常规免疫报表数据,采用指数平滑模型、ARIMA模型对2012—2018年数据建立预测模型,使用2019年数据验证模型的预测效果。结果使用2012—2018年数据建立的预测模型两种时间序列分析法的拟合精度指标平稳R^(2)分别为0.751和0.459,均方根误差(RMSE)分别为551.521和785.371,平均绝对百分误差(MAPE)分别为9.284%和11.856%,平均绝对误差(MAE)分别为424.191和537.272,使用2019年数据验证的平均相对误差(MRD)分别为12.495%和12.572%。结论指数平滑法对预防接种门诊接种量具有很好的预测效果,可以为预防接种门诊管理提供预案和决策依据。Objective To compare the accuracy of exponential smoothing prediction model and ARIMA model in the prediction of vaccine dose administered in immunization clinics.Methods Collecting the data of dose administered by month in Xihongmen Immunization Clinics of Beijing from 2012 to 2019.Exponential smoothing prediction model and ARIMA model were constructed to predicate the data from 2012 to 2018.The actual data in 2019 were used to confirm the predicted results.Results R^(2)of two models were 0.751 and 0.459.RMSE were 551.521 and 785.371.MAPE were 9.284%and 11.856%.MAE were 424.191 and 537.272.MRD were 12.459%and 12.572%.Conclusion The exponential smoothing method is better than the ARIMA method for predicting vaccine dose administered in immunization clinics,so it can be used to provide references for immunization clinics management and decision-making.

关 键 词：指数平滑法 ARIMA模型 预防接种 

分 类 号：R197.1[医药卫生—卫生事业管理]