时间序列分析的直播星用户开卡预测
|
摘要
针对同一地区不同时间内直播星电视服务用户的数量变化特征,该文提出了基于时间序列的直播星用户预测方法,并通过ARIMA模型预测用户数量。分析了ARIMA模型的建立方法和评价参数,对时间序列模型ARIMA(p,d,q)的参数进行不同组合的尝试;利用模型评价参数(平稳的R方和正态化的BIC)确定模型的阶数p、q;对模型进行参数检验和显著性检验,根据检验结果确立最终模型为ARIMA(4,2,3);采用宁夏用户数据对模型拟合效果进行验证,进而对未来几个月的用户数量进行预测。结果表明,时间序列模型ARIMA(4,2,3)对宁夏用户数量变化的预测准确度比较高。
According to the quantitative variation characteristics of live satellite users at different time in the same area,this paper proposed a prediction method of users based on time series and predicted the users' number by ARIMA model.The method of establishing ARIMA model and evaluation parameters were analyzed,and different combinations to the parameters of ARIMA(p,d,q)of time series model were tried;the model evaluation parameters(stationary R-squared and normalized BIC)were used to determine a model for the order of pand q;the model parameters and significance were tested,and the model ARIMA(4,2,3)was established according to test results.The ARIMA model of time series was applied to predict the number of users for the next few months in Ningxia,and the results showed that its prediction accuracy was relatively high.
According to the quantitative variation characteristics of live satellite users at different time in the same area,this paper proposed a prediction method of users based on time series and predicted the users' number by ARIMA model.The method of establishing ARIMA model and evaluation parameters were analyzed,and different combinations to the parameters of ARIMA(p,d,q)of time series model were tried;the model evaluation parameters(stationary R-squared and normalized BIC)were used to determine a model for the order of pand q;the model parameters and significance were tested,and the model ARIMA(4,2,3)was established according to test results.The ARIMA model of time series was applied to predict the number of users for the next few months in Ningxia,and the results showed that its prediction accuracy was relatively high.
引文
[1]张红玉,杨飞华,李国学,等.基于ARIMA模型的北京市朝阳区建筑垃圾产量分析与预测[J].环境工程,2014,32(S1):696-699.
[2]NASIBOVE,PEKER S.Time series labeling algorithms based on the K-nearest neighbors’frequencies[J].Expert Systems with Application,2011,38(5):5028-5035.
[3]魏宁,边宽江,袁志发.基于ARIMA模型的陕西省GDP分析与预测[J].安徽农业科学,2010,38(9):4933-4935.
[4]李生彪,彭建奎.基于Box-Jenkins方法的甘肃省GDP时间序列分析建模与预测[J].延边大学学报:自然科学版,2014,40(2):146-149.
[5]彭斯俊,沈加超,朱雪.基于ARIMA模型的PM2.5预测[J].安全与环境工程,2014,21(6):125-128.
[6]石子泊,邹志红.基于小波变换的ARIMA模型在水质预测中的应用研究[J].环境工程学报,2014,8(10):4550-4554.
[7]孙苗,孙祥超,耿伟华.基于ARIMA模型的山东省月降水量时间序列分析[J].鲁东大学学报:自然科学版,2013,29(3):244-249.
[8]虞安,王忠.基于ARIMA模型与时间序列的城市旅游倾向预测[J].统计与决策,2014(13):86-89.
[9]张侨,蔡道成.基于ARIMA模型的海南旅游者人数预测[J].科技和产业,2014(4):24-27.
[10]SHAHANDASHTI S M,SHURI B A.Forecasting engineering news-record construction cost index using multivariate time series models[J].Journal of Construction Engineering and Management,2013,139(9):615-617.
[11]刘劲松.数据挖掘中的现代时间序列分析方法[J].信息技术,2007,1(7):100-101.
[12]乐天,蔡远文,马雪松.时间序列预测的发展与应用[J].兵工自动化,2015,34(2):63-68.
[13]刘薇.时间序列分析在吉林GDP预测中的应用[D].长春:东北师范大学,2008:5.
[14]李大利.一种时间序列预测方法的应用研究[D].长春:吉林大学,2008:2.
[15]魏宁.时间序列分析方法研究及其在陕西省GDP预测中的应用[D].陕西:西北农林科技大学,2010:1.
[16]刘荣其.基于聚类和时序分析的林分生长模型[D].福州:福建农林大学,2009:11.
[2]NASIBOVE,PEKER S.Time series labeling algorithms based on the K-nearest neighbors’frequencies[J].Expert Systems with Application,2011,38(5):5028-5035.
[3]魏宁,边宽江,袁志发.基于ARIMA模型的陕西省GDP分析与预测[J].安徽农业科学,2010,38(9):4933-4935.
[4]李生彪,彭建奎.基于Box-Jenkins方法的甘肃省GDP时间序列分析建模与预测[J].延边大学学报:自然科学版,2014,40(2):146-149.
[5]彭斯俊,沈加超,朱雪.基于ARIMA模型的PM2.5预测[J].安全与环境工程,2014,21(6):125-128.
[6]石子泊,邹志红.基于小波变换的ARIMA模型在水质预测中的应用研究[J].环境工程学报,2014,8(10):4550-4554.
[7]孙苗,孙祥超,耿伟华.基于ARIMA模型的山东省月降水量时间序列分析[J].鲁东大学学报:自然科学版,2013,29(3):244-249.
[8]虞安,王忠.基于ARIMA模型与时间序列的城市旅游倾向预测[J].统计与决策,2014(13):86-89.
[9]张侨,蔡道成.基于ARIMA模型的海南旅游者人数预测[J].科技和产业,2014(4):24-27.
[10]SHAHANDASHTI S M,SHURI B A.Forecasting engineering news-record construction cost index using multivariate time series models[J].Journal of Construction Engineering and Management,2013,139(9):615-617.
[11]刘劲松.数据挖掘中的现代时间序列分析方法[J].信息技术,2007,1(7):100-101.
[12]乐天,蔡远文,马雪松.时间序列预测的发展与应用[J].兵工自动化,2015,34(2):63-68.
[13]刘薇.时间序列分析在吉林GDP预测中的应用[D].长春:东北师范大学,2008:5.
[14]李大利.一种时间序列预测方法的应用研究[D].长春:吉林大学,2008:2.
[15]魏宁.时间序列分析方法研究及其在陕西省GDP预测中的应用[D].陕西:西北农林科技大学,2010:1.
[16]刘荣其.基于聚类和时序分析的林分生长模型[D].福州:福建农林大学,2009:11.