基于ARMA-TS-GARCH有限混合模型的交通数据分析  

Analysis of traffic data based on ARMA-TS-GARCH finite mixture model

作　　者：王维强[1] 牛振东[1] 曹玉娟[1] 赵育民[1] 赵堃[1] 

机构地区：[1]北京理工大学计算机学院,北京100081

出　　处：《中南大学学报（自然科学版）》2010年第5期1860-1864,共5页Journal of Central South University:Science and Technology

基　　金：国家自然科学基金资助项目(60803050)

摘　　要：基于对时间序列数据进行研究时不仅需要对它们的自回归性和周期性进行分析,而且需要对序列的方差与长期性无规则波动进行分析等问题,提出基于ARMA模型和APARCH模型的ARMA-TS-GARCH模型,并且对该模型的构造进行分析,针对洛杉矶长滩地区交通量数据进行模型参数估计和诊断检验,将此模型与GARCH和ARMA-GARCH模型进行比较。研究结果表明:ARMA-TS-GARCH模型对数据的拟合要优于ARMA模型和APARCH模型;用ARMA-TS-GARCH模型对数据集进行预测,所得结果较理想。Based on the fact that in order to study time-series data,their auto-regression and periodic and the variance of the sequence and long-term random fluctuations must be analyzed,ARMA-TS-GARCH model based on ARMA model and APARCH model was proposed,and the construction of the model and the parameter estimation with the Los Angeles Long Beach area traffic data and diagnostic tests were analyzed.The results of this model were compared with those of the GARCH and ARMA-GARCH model.The results show that the ARMA-TS-GARCH model fits the data better than ARMA model and APARCH model.Using ARMA-TS-GARCH model to predict the traffic data set,the results are well.

关 键 词：ARMA-TS-GARCH模型 时间序列 预测 交通 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]