ADCINAR(1)模型的加权条件最小二乘估计  

Weighed Conditional Least Squares Estimation for ADCINAR(1) Model

作　　者：王宇 王纯杰 张海祥 WANG Yu;WANG Chunjie;ZHANG Haixiang(Center for Applied Mathematics,Tianjin University,Tianjin 300072,China;School of Mathematics and Statistics,Changchun University of Technology,Changchun 130012,China)

机构地区：[1]天津大学应用数学中心,天津300072 [2]长春工业大学数学与统计学院,长春130012

出　　处：《吉林大学学报（理学版）》2019年第3期553-561,共9页Journal of Jilin University:Science Edition

基　　金：国家自然科学基金(批准号:11671054);天津大学北洋学者.青年骨干教师计划项目(批准号:2018XRG-0038)

摘　　要：用加权条件最小二乘方法,对基于相依计数序列的一阶整值自回归模型(ADCINAR(1))进行参数估计,给出参数估计的表达式及其渐近分布,并推导模型的高阶矩、高阶累积量、谱密度和双谱密度.数值模拟结果表明,将加权条件最小二乘估计、条件最小二乘估计和Yule-Walker估计进行比较,验证了加权条件最小二乘方法的有效性.Using the weighed conditional least squares method, we estimated the parameters of the first-order integer-valued autoregressive model (ADCINAR(1)) based on dependent counting series. We gave the expression of parameter estimation and its asymptotic distribution, and derived the higher-order moments, higher-order cumulants, spectral density and bispectral density of the model. The numerical simulation results show that the validity of weighed conditional least squares method is verified by comparing weighed conditional least squares estimation with conditional least squares estimation and Yule-Walker estimation.

关 键 词：双谱密度 相依计数序列 高阶矩 整值时间序列 加权条件最小二乘 

分 类 号：O212.7[理学—概率论与数理统计]