抗差递推校正模型的风险分析

Abstract

抗差递推校正模型在抵御流量异常值对实时校正精度影响方面效果明显,但同时也存在将真实值作为异常值剔除,校正误差更大的风险。利用蒙特卡洛方法,分析不同状况下,抗差递推校正模型风险与效果的关系。结果表明抗差递推校正模型的风险受异常值发生频率影响较大。

Robust recursive updating model is insensitive to the outliers and is effective to stable flood updating accuracy.At the same time,it is risky to detect falsely good value as outliers.Based on Monte Carlo Method,the relations between risk and effect of model are obtained.The research results indicate that the risk of the model is affected by frequency of outliers.