基于求和自回归滑动平均模型的道路交通伤害死亡趋势预测分析

寻鲁宁; 张帆; 孙纪新; 曹亚景; 孙祯; 史卫卫; 李玫; 崔泽

doi:10.16462/j.cnki.zhjbkz.2020.04.019

基于求和自回归滑动平均模型的道路交通伤害死亡趋势预测分析

doi: 10.16462/j.cnki.zhjbkz.2020.04.019

寻鲁宁¹,
张帆²,
孙纪新²,
曹亚景²,
孙祯¹,
史卫卫²,
李玫²,
崔泽^1, ,

1.
063210 唐山, 华北理工大学公共卫生学院
2.
050011 石家庄, 河北省疾病预防控制中心慢性非传染性疾病防治所

基金项目: 2017年度河北省医学科学研究重点课题计划(20170447)

详细信息

通讯作者:
崔泽, E-mail:cuize11@163.com

中图分类号: R181.2;U491.31
计量
- 文章访问数: 350
- HTML全文浏览量: 153
- PDF下载量: 42
- 被引次数: 0
出版历程
- 收稿日期: 2019-11-08
- 修回日期: 2020-02-03
- 刊出日期: 2020-04-10

Prediction and analysis of road traffic injury death trend based on ARIMA model

1.
College of Public Health, North China University of Science and Technology, Tangshan 063210, China
2.
Department of Chronic Non-communicable Disease Control and Prevention, Hebei Provincial Center for Disease Control and Prevention, Shijiazhuang 050011, China

Funds: 2017 Hebei Provincial Medical Science Research Key Project Plan(20170447)

More Information

Corresponding author: CUI Ze, E-mail:cuize11@163.com

摘要

摘要: 目的分析2014-2018年河北省道路交通伤害死亡情况, 探讨求和自回归滑动平均模型(autoregressive integrated moving average model, ARIMA)在道路交通伤害死亡趋势预测中的可行性。方法采用描述流行病学分析2014-2018年河北省道路交通伤害死亡概况, 运用R 3.5.3软件对河北省2014年1月-2018年6月道路交通伤害月度死亡资料建立ARIMA预测模型, 进行整体回代观察拟合效果, 比较2018年7月-12月预测值与真实值, 评价预测效果。结果 2014-2018年河北省累计报告道路交通伤害死亡人数13 147例, 男性10 071例, 女性3 076例, 年均死亡率为17.79/10万, 总体呈现下降趋势。构建的最佳预测模型为ARIMA(0, 1, 1)(0, 1, 1)₁₂, 赤池信息量准则(Akaike information criterion, AIC)为390.64, Schwaz贝叶斯准则(Schwarz Bayesian criterion, SBC)为395.78;残差序列为白噪声序列(均有P > 0.05), 模型参数非零(均有P < 0.05);预测结果实际值均落在预测值95%置信区间内, 预测值与实际值之间的相对误差在1.15%~11.85%之间, RMSE=13.65, MAE=10.88, MAPE=4.80%, 模型预测性能良好。结论河北省道路交通伤害死亡水平总体呈逐年下降趋势, ARIMA模型可用于道路交通伤害死亡趋势的短期预测。
- 求和自回归滑动平均模型 /
- 道路交通伤害 /
- 死亡趋势 /
- 预测
Abstract: Objective To analyze the death trend of road traffic injury in Hebei province from 2014 to 2018, and to discuss the feasibility of autoregressive integrated moving average model(ARIMA) in the prediction of road traffic injury deaths trend. Methods Descriptive epidemiological method was used to analyze the general situation of road traffic injury deaths in Hebei province from 2014 to 2018. R 3.5.3 was used to establish the ARIMA prediction model for the monthly death cases of road traffic injury in Hebei province from January 2014 to June 2018. The overall regression was used to observe the fitting effect, the predicted value and the real value were compared from July to December in 2018 to evaluate the prediction effect. Results A total of 13 147 road traffic injury deaths from 2014 to 2018 were reported by Hebei province. The number of road traffic injury deaths was 10 071 males and 3 076 females, with an annual mortality rate of 17.79/100 000, showing a downward trend overall. The best prediction model is ARIMA(0, 1, 1)(0, 1, 1)₁₂. Akaike information criterion(AIC) is 390.64, Schwarz Bayesian criterion(SBC) is 395.78, the residual sequence is white noise sequence(P > 0.05), and the parameters of the model are significantly non-zero(P < 0.05). The actual values of the prediction results all fall within the 95% confidence interval of the predicted values, and the relative error between the predicted values and the actual values is between 1.15% and 11.85%. The root mean square error(RMSE) is 13.65, the mean absolute error(MAE) is 10.88, and the mean average percentage error(MAPE) is 4.80%. The prediction performance of the model is good. Conclusions The overall road traffic injury deaths in Hebei province show a downward trend year by year. ARIMA model can be used to predict the short-term trend of road traffic injury deaths.
- ARIMA model /
- Road traffic injury /
- Death trend /
- Prediction

HTML全文

图 1 2014-2018年河北省道路交通伤害死亡率变化情况

Figure 1. Annual mortality rate of road traffic injury in Hebei from 2014 to 2018

下载: 全尺寸图片幻灯片

图 2 时间序列分解图

Figure 2. Decomposition diagram of time series

下载: 全尺寸图片幻灯片

图 3 差分后序列自相关图

Figure 3. Autocorrelation diagram after difference

下载: 全尺寸图片幻灯片

图 4 差分后序列偏自相关图

Figure 4. Partial autocorrelation diagram after difference

下载: 全尺寸图片幻灯片

图 5 ARIMA(0, 1, 1)(0, 1, 1)₁₂拟合回代曲线

Figure 5. ARIMA(0, 1, 1)(0, 1, 1)₁₂ fitting regression curve

下载: 全尺寸图片幻灯片

表 1 ARIMA模型拟合效果比较

Table 1. Comparison of fitting effects of ARIMA model

模型	AIC	SBC	Log likelihood
ARIMA(0, 1, 1)(0, 1, 1)₁₂	390.64	395.78	-192.32
ARIMA(0, 1, 1)(0, 1, 2)₁₂	391.46	398.31	-191.73
ARIMA(0, 1, 1)(1, 1, 0)₁₂	391.82	396.96	-192.91
ARIMA(0, 1, 1)(1, 1, 1)₁₂	391.54	398.39	-191.77

下载: 导出CSV

表 2 ARIMA(0, 1, 1)(0, 1, 1)₁₂模型的参数检验

Table 2. Parameter test of ARIMA(0, 1, 1)(0, 1, 1)₁₂ model

参数	估计值	S_x值	t值	P值
非季节部分
MA1	-1.00	0.11	-8.95	< 0.001
季节部分
SMA1	-0.45	0.27	-1.68	0.049

下载: 导出CSV

表 3 2018年7月-2018年12月道路交通伤害死亡例数预测值与实际值比较

Table 3. Comparison between predicted value and actual value of road traffic injury death cases from July 2018 to December 2018

月份	实际值	预测值	95% CI			绝对误差	相对误差(%)
月份	实际值	预测值	下限		上限	绝对误差	相对误差(%)
7	218	192.18	143.93	240.43		25.82	11.85
8	207	210.57	162.32	258.82		-3.57	1.72
9	245	229.61	181.36	277.86		15.39	6.28
10	236	230.59	182.34	278.84		5.41	2.29
11	236	248.95	200.70	297.20		-12.95	5.49
12	187	189.16	140.91	237.41		-2.16	1.15
合计	1 329	1 301.06	1 011.56	1 590.55		27.94	2.10

下载: 导出CSV

参考文献(19)

[1]	World Health Organization. Road traffic injuries[EB/OL]. (2018-01-16)[2018-05-07]. http://www.who.int/violence_injury_prevention/road_traffic/en.
[2]	Zhang X, Xiang H, Jing R, et al. Road traffic injuries in the people's republic of China, 1951-2008[J]. Traffic Inj Prev, 2011, 12(6): 614-620. DOI: 10.1080/15389588.2011.609925.
[3]	袁慧, 王声湧.我国伤害预防与控制工作的主要进展及展望[J].中华疾病控制杂志, 2017, 21(10): 971-973, 978. DOI: 10.16462/j.cnki.zhjbkz.2017.10.001. Yuan H, Wang SY. Progress and prospect on injury prevention and control in China[J]. Chin J Dis Control Prev, 2017, 21(10): 971-973, 978. DOI: 10.16462/j.cnki.zhjbkz.2017.10.001.
[4]	庞媛媛, 张徐军, 崔梦晶, 等.道路交通伤害预测方法研究进展[J].伤害医学(电子版), 2013, 2(2): 49-53. DOI: 10.3868/j.issn.2095-1566.2013.02.011. Pang YY, Zhang XJ, Cui MJ, et al. Research progresses of predicting methods of road traffic injuries[J]. Injury Medicine(Electronic Edition), 2013, 2(2): 49-53. DOI: 10.3868/j.issn.2095-1566.2013.02.011.
[5]	高刘伟, 张徐军, 周义夕, 等. BP神经网络在道路交通伤害预测应用的研究进展[J].伤害医学(电子版), 2019, 8(1): 53-57. DOI: 10.3868/j.issn.2095-1566.2019.01.010. Gao LW, Zhang XJ, Zhou YX, et al. Research progress of BP neural network in prediction of road traffic injury[J]. Injury Medicine(Electronic Edition), 2019, 8(1): 53-57. DOI: 10.3868/j.issn.2095-1566.2019.01.010.
[6]	尹锡玲, 代文灿, 李德云, 等. 2004-2016年珠海市道路交通伤害时间序列分析[J].实用预防医学, 2019, 26(5): 555-558. DOI: 10.3969/j.issn.1006-3110.2019.05.012. Yin XL, Dai WC, Li DY, et al. Time series analysis of road traffic injuries in Zhuhai city, 2004-2016[J]. Pract Prevent Med, 2019, 26(5): 555-558. DOI: 10.3969/j.issn.1006-3110.2019.05.012.
[7]	庞媛媛, 张徐军, 涂志斌, 等.自回归移动平均混合模型在中国道路交通伤害预测中的应用[J].中华流行病学杂志, 2013, 34(7): 736-739. DOI: 10.3760/cma.j.issn.0254-6450.2013.07.018. Pang YY, Zhang XJ, Tu ZB, et al. Autoregressive integrated moving average model in predicting road traffic injury in China[J]. Chin J Epidemiol, 2013, 34(7): 736-739. DOI: 10.3760/cma.j.issn.0254-6450.2013.07.018.
[8]	高景宏, 朱瑶, 熊黎黎, 等.汕头市某三甲医院2002-2012年交通伤害病例的时间序列分析[J].中华疾病控制杂志, 2014, 18(10): 917-921. http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=jbkzzz201410003 Gao JH, Zhu Y, Xiong LL, et al. The traffic injuries of a tertiary hospital in Shantou city: a time-series analysis, 2002-2012[J]. Chin J Dis Control Prev, 2014, 18(10): 917-921. http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=jbkzzz201410003
[9]	王燕.时间序列分析-基于R[M].北京: 中国人民大学出版社, 2015. Wang Y. Time series analysis with R[M]. Beijing: China Renmin University Press, 2015.
[10]	洪志敏, 郝慧, 房祥忠, 等. ARIMA模型在京津冀区域手足口病发病趋势预测中的应用[J].数理统计与管理, 2018, 37(2): 191-197. DOI: 10.13860/j.cnki.sltj.20171013-002. Hong ZM, Hao H, Fang XZ, et al. Application of ARIMA model in predicting the incidence trend of hang, foot and mouth disease in Beijing, Tianjin and Hebei[J]. Journal of Applied Statistics and Management, 2018, 37(2): 191-197. DOI: 10.13860/j.cnki.sltj.20171013-002.
[11]	朱佳佳, 胡登利, 洪秀琴, 等.基于时空大数据的甲型肝炎发病率分布特征分析及预测模型[J].中华疾病控制杂志, 2018, 22(11): 1144-1147. DOI: 10.16462/j.cnki.zhjbkz.2018.11.012. Zhu JJ, Hu DL, Hong XQ, et al. Analysis of distribution characteristics and prediction model of hepatitis a incidence based on spatiotemporal big data[J]. Chin J Dis Control Prev, 2018, 22(11): 1144-1147. DOI: 10.16462/j.cnki.zhjbkz.2018.11.012.
[12]	孟凡东, 吴迪, 隋承光. 2004-2015年中国狂犬病发病数据ARIMA乘积季节模型的建立及预测[J].中国卫生统计, 2016, 33(3): 389-391, 395. http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=zgwstj201603006 Meng FD, Wu D, Sui CG. Human rabies incidence in China: trends and predictions from a time series analysis from 2004 through 2015[J]. Chin J Health Statistics, 2016, 33(3): 389-391, 395. http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=zgwstj201603006
[13]	Wang X, Yu H, Nie C, et al. Road traffic injuries in China from 2007 to 2016: the epidemiological characteristics, trends and influencing factors[J]. Peer J, 2019, 7: e7423. DOI: 10.7717/peerj.7423.
[14]	Leilei D, Pengpeng Y, Haagsma JA, et al. The burden of injury in China, 1990-2017: findings from the global burden of disease study 2017[J]. Lancet Public Health, 2019, 4(9): 449-461. DOI: 10.1016/S2468-2667(19)30125-2.
[15]	栗华, 朱硕斌, 张中朝, 等. 2006-2009年河北省部分城乡居民道路交通伤害监测分析[J].中国慢性病预防与控制, 2011, 19(3): 257-259. http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=zgmxbyfykz201103014 Li H, Zhu SB, Zhang ZC, et al. Analysis of road traffic injuries among the residents of some urban and rural areas Hebei province from 2006 to 2009[J]. Chin J Prev Contr Chron Dis, 2011, 19(3): 257-259. http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=zgmxbyfykz201103014
[16]	刘莹, 胡锡敏, 陈言, 等ARIMA模型在海南省2014-2017年居民自杀死亡率中的应用[J].中华流行病学杂志, 2018, 39(5): 664-668. DOI: 10.3760/cma.j.issn.0254-6450.2018.05.024. Liu Y, Hu XM, Chen Y, et al. Application of ARIMA model in prediction of mortality rate of suicide in Hainan province[J]. Chin J Epidemiol, 2018, 39(5): 664-668. DOI: 10.3760/cma.j.issn.0254-6450.2018.05.024.
[17]	王晨, 郭倩, 周罗晶.基于R语言的ARIMA模型对流感样病例发病趋势的预测[J].中华疾病控制杂志, 2018, 22(9): 957-960. DOI: 10.16462/j.cnki.zhjbkz.2018.09.020. Wang C, Guo Q, Zhou LJ. Forecast of incidence trend of influenza-like illness by the ARIMA model based on R[J]. Chin J Dis Control Prev, 2018, 22(9): 957-960. DOI: 10.16462/j.cnki.zhjbkz.2018.09.020.
[18]	杨慧欣, 赵晨皓, 雒静静, 等. 2011-2018年我国登革热疫情时间序列分析及空间自相关分析[J].中华疾病控制杂志, 2019, 23(10): 1250-1254. DOI: 10.16462/j.cnki.zhjbkz.2019.10.018. Yang HX, Zhao CH, Luo JJ, et al. Time series anlysis and spatial autocorrelation analysis of dengue data in China from 2011 to 2018[J]. Chin J Dis Control Prev, 2019, 23(10): 1250-1254. DOI: 10.16462/j.cnki.zhjbkz.2019.10.018.
[19]	王克伟, 李金平, 邓超, 等.细菌性痢疾自回归滑动平均和非线性自回归组合模型预测研究[J].第二军医大学学报, 2017, 38(10): 1315-1320. DOI: 10.16781/j.0258-879x.2017.10.1315. Wang KW, Li JP, Deng C, et al. Application of ARIMA-NAR combined model in predicting bacillary dydentery[J]. Acad J Sec Mil Med Univ, 2017, 38(10): 1315-1320. DOI: 10.16781/j.0258-879x.2017.10.1315.