利用偏度分析建立动态检测门限的野值剔除方法

提出了一种采用动态门限检测和剔除野值的新方法。以往的单一检测门限不能可靠地清除观测数据中的全部野值 ,原因是缺少判断有无未剔除野值的识别机制。这里 ,借助于统计偏度分析建立了这种判别机制 ,从而能够有效地判断出是否存在有未被剔除的野值 ,并通过调节野值检测门限来达到彻底清除野值的目的。最后 ,还将此法推广到复数域

Dynamic Threshold Detection for Wild Value Elimination Based on Skewness Analysis

This paper presents a new wild-value-eliminated method using a dynamic detection threshold. With the single threshold detection the wild values in observed data can not be cleared out reliably due to no recognition function for judging whether the uncancelled wild values exist or not. Here, a recognition method is made by the statistical skewness analysis. It can effectively judge the uncancelled wild values existing and then clear out them by adjusting the wild detecting threshold. Finally, this method is also applied to the complex data.