| ARIMA模型在预测我国乙肝月报告发病数中的应用 |
| |
| 引用本文： | 沈忠周,王雅文,马帅,赵鹏宇,于凯,严宝湖,江宇.ARIMA模型在预测我国乙肝月报告发病数中的应用[J].中华疾病控制杂志,2018,22(3):298. |
| |
| 作者姓名： | 沈忠周 王雅文 马帅 赵鹏宇 于凯 严宝湖 江宇 |
| |
| 作者单位： | 北京协和医学院公共卫生学院流行病与卫生统计学系, 北京 100730 |
| |
| 基金项目： | 北京协和医学院青年科研基金资助（2017330021） |
| |
| 摘 要： | 目的 建立适合预测我国乙肝月报告发病人数的自回归移动平均（autoregressive integrated moving average，ARIMA）模型。方法 收集2010年3月~2017年8月我国乙肝月报告发病人数据资料，用Excel 2010建立数据库，用R 3.3.3软件进行模型构建。其中2010年3月~2017年3月数据用于模型建立，2017年4~8月数据用于模型检验。结果 我国乙肝月报告发病数整体呈现下降趋势并于每年2月达到最低值，于3月份迅速上升到高峰值，具有明显季节性和周期性。建立ARIMA（2，1，1）（1，1，1）12模型对我国乙肝月报告发病数进行预测，该模型预测的绝对误差平均值为2 628.55，相对误差最大值为6.16%，最小值为1.29%，平均值为2.61%。结论 基于本研究数据，ARIMA（2，1，1）（1，1，1）12模型能较好地拟合我国乙肝的月报告发病人数，可用于预测。
|
| 关 键 词： | 肝炎 乙肝 模型 统计学 预测 |
| 收稿时间： | 2017-09-04 |
Prediction on the number of hepatitis B by ARIMA model in China |
| |
| Affiliation: | Department of Epidemiology and Biostatistics, School of Public Health, Peking Union Medical College, Beijing 100730, China |
| |
| Abstract: | Objective To develop a suitable autoregressive integrated moving average (ARIMA) model for predicting the new reported number of hepatitis B in China. Methods The data was collected from March 2010 to August 2017 to built the database by Excel 2010. The model was built by R 3.3.3 software with data from March 2010 to March 2017 and the model's predicting ability was evaluated by the data from April 2017 to August 2017. Results A strong seasonal and decreasing nature, the lowest value in February and the highest peak in March, were observed in the new reported cases of hepatitis B. The ARIMA(2,1,1)(1,1,1)12 model was chosen to forecast the new report cases of hepatitis B in China. By comparing the data, it showed that the mean absolute error, the maximum relative error, the minimum relative error and the mean relative error of this model were 2 628.55, 6.16%, 1.29% and 2.61%, respectively. Conclusions The ARIMA (2,1,1)(1,1,1)12 model is suggested to give a good prediction on the monthly number of hepatitis B in China. |
| |
| Keywords: | |
|
| 点击此处可从《中华疾病控制杂志》浏览原始摘要信息 |
|
点击此处可从《中华疾病控制杂志》下载全文 |
|
