包含异常数据的居民出行稳健回归分析
详细信息 本馆镜像全文    |  推荐本文 | | 获取馆网全文
摘要
采用稳健回归方法克服了最小二乘估计受异常样本点影响较大的弊病,该方法通过对不同的点施加不同的权重减少"异常点"作用,并据此建立加权的最小二乘估计,反复迭代以改进权重系数,直至权重系数之改变小于一定的允许误差,得到模型参数更加贴近实际值.案例研究表明,用稳健回归方法建立的数学模型避免了少数异常值干扰的影响,更加真实地反映了居民出行发生的变化趋势.
This paper adopts robust regression to overcome the impacts that abnormal sample points have on least squares method,and model parameters are closer to reality.Actual case shows that the mathematical model established by using the robust regression method avoids the interference effects of some anomalous values,more truly reflects the changing trend of the residents travel and is a powerful tool in analyzing the residents travel trend.
引文
[1]Ortiz M C.Robust regression techniques a useful al-ternative for the detection of outlier data in chemicalanalysis[J].Talanta,2006,70(3):499-512.
    [2]Pennacchi P.Robust estimate of excitations in me-chanical systems using m-estimators-theoretical back-ground and numerical applications[J].Journal ofSound and Vibration,2008,310:923-946.
    [3]杨马陵.地震前兆数据的稳健回归与建模[J].西北地震学报,1999(12):399-408.
    [4]周蒂.稳健统计学与地球化学数据的统计分析[J].地球科学,1991,16(3):273-279.
    [5]Hample F R.Robust statistics:the approach basedon influence function[M].New York:John Wiley&Sons,1986.
    [6]Huber P.Robust Statistics[M].New York:JohnWiley,1981.

版权所有:© 2023 中国地质图书馆 中国地质调查局地学文献中心