CAJ | 학술논문

运用局部影响分析对二维AR（1）模型进行了统计诊断，克服了数据删除对时间序列样本数据相依性的破坏，得出了影响曲率的具体计算公式，最后运用实例说明了该方法的有效性。本文的方法可以一次性探测出多维时间序列模型中的强影响点，提高了寻找模型强影响点的效率。
운용국부영향분석대이유AR（1）모형진행료통계진단，극복료수거산제대시간서렬양본수거상의성적파배，득출료영향곡솔적구체계산공식，최후운용실례설명료해방법적유효성。본문적방법가이일차성탐측출다유시간서렬모형중적강영향점，제고료심조모형강영향점적효솔。
To find the outliers and influence points in the two dimensional AR(1 )models,the curvature formula of local in-fluence is calculated which help to detect all of the influence points of the models in one time.In the end,a numerical exam-ple is given to illustrate the effectiveness of the conclusion.With this method,influence cases in the multi-dimensional time series model is detected once.The efficiency of the influence points search is greatly improved.