异常值

该算法主要的特点就是利用了K-MEANS聚类算法原理上的缺陷，即会陷入局部极小的特点，在基于聚类的异常值检测的思想下，以聚类的方式寻找异常值并将其删除。

The main point is that the defect that K-MEANS would fall into a suboptimization is made advantage in our algorithm . Under the strategy of cluster-based outlier detection , outliers are searched and deleted in clusters .

Gillis先生表示，投资者应当正确审视所取得的业绩。“总体来说，业绩确实让人震惊，盈利高达100亿美元。但是从很多指标来说，都是异常值，因此应当审视相对于预期的业绩表现。”

However , Mr Gillis said that investors should put the results into perspective . " Overall the results are stunning - it 's made $ 10bn in profit . But Apple is an outlier in many metrics , so you need to look at the performance relative to expectations , " he said .

用Origin剔除线性拟合中实验数据的异常值

Data Rejection in Linear Fitting Using Origin

金属薄板塑性应变比r值异常值的处理

Treatment of Abnormal Value of Plastic Strain Ratio r Value about Metal Thin Steel Plate

Logistic回归模型异常值的判定原理探讨

The principled research to assess the outliers of Logistic regression model

再用较少已测点的原始数据，逼近GPS水准网中较多检核点的高程异常值。

Then , use few known point of original data to approximate the large number of check point in GPS network .

在指数样本异常值的检验中，Fisher型统计量是十分有效的。

The Fisher-type statistic is very effective for testing outlier in exponential sample .

双向分类随机可加效应模型中可加性异常值（outlier）的检验

The test about the additivity outlier in the two-way classification stochastic model

主成分分析方法由于使用了L2范数，因此对异常值较敏感。

Because of using L2 norm , Principal Component Analysis ( PCA ) method is sensitive to outliers .

这样双参数Weibull分布场合下异常值的问题得到很好地解决。

As a result , the problem of testing the outliers in two-parameter Weibull examples is solved primely .

主要工作和创新点如下：1、基于不同类型识别变量的异常值探测的Bayes方法。

The main conclusions areas follows : 1 . Bayes method for outlier detection based on different types of classification variables .

笔者综合运用非线性回归诊断分析，建立了Logistic回归模型异常值判定的基本原理和方法。

On the basis of the non-linear regression diagnostic analysis , the authors put forward a diagnostic method to assess the outliers of the Logistic regression model .

Ⅰ型极值分布样本异常值的Irwin型检验

Irwin type test for outliers in the sample of type I extreme value distribution

对该方法进行了计算机MonteCarlo模拟比较，结果显示当异常值存在时，该法优于传统的Fisher判别方法。

Results of Monte Carlo simulation and practical examples showed that the new procedure wins advantages over the classical Fisher discrimination when some outliers exist in data set .

数据序列中少量异常值（Outliers）的存在会严重破坏传统最优化估计方法的估值性能，这时须采用韧性估计方法才有效。

Outliers in time series can play havoc with conventional optimizing estimation procedures , in which cases the robust method must be used .

本文给出用于指数样本中检验最大（或最小）的观察值是否为异常值的Fisher型统计量分布的分位点。

In this paper , the percentile points are given for Fisher-Type statistics for testing the largest or smallest observational value as outlier in exponential sample .

正态样本异常值Nair检验统计量的近似分布

The approximate distribution of Nair test statistics of detecting Outlier for normal sample

在有异常值的数据中，Bootstrap样本可能比原有样本含有更高的污染，这会降低所要做的统计推断的有效性。

For the data with outliers , Bootstrap samples may contain more " pollution " than original samples and reduce the validity of our statistical deduction .

该文利用支持向量回归算法中结构风险函数的性质以及KT条件，提出一种回归中的异常值检测方法。

A method of outlier detection in re-gression is proposed making use of the character of structure risk function and KT condition in support vector regres-sion in this paper .

在JX地区，由莫氏界面的起伏只能引起21&42毫伽的重力异常值。

The undulation of Mohorovicic discontinuity under JX region only causes the gravity anomaly of 21-42 milligals .

利用小波变换对BOTDR的测量数据进行消噪处理和异常值检测，并用某工程现场实测的BOTDR数据进行了验证。

Denoising and singular point detecting of the measured signal is performed using wavelet analysis .

而后，根据以上计算原理建立了搜索最优方程的计算流程，采用这一计算流程可以查找异常值，然后利用EM算法对异常值进行了调整。

The computation flow for finding the best fitting equation is founded by LSE , M-estimation , R-estimation and successive regression , abnormal values can be found by the computation flow and adjusted by EM-algorithm .

卵巢癌术后复发者血清IAP水平明显高于健康妇女IAP水平（P＜0.01），异常值的发生率为100%。

In patients with recurrence of ovary tumor , the IAP levels was increased to compare with the health women ( P < 0.01 ), the incidence of the abnormal value was 100 % .

然后探究高斯混合模型、聚类EM算法的原理，并对ARX模型残差序列进行聚类分析和异常值检测。

Finally , we analyze the principle of Gaussian mixture model and clustering EM algorithm , which should be applying in clustering analysis and anomaly detection of the ARX model outlier series .

当流量资料存在异常值，对AR模型参数分别采用递推最小二乘估计和抗差递推最小二乘估计进行了分析计算。

For outliers occurring in the measured data of flood discharge , the parameters of the AR model obtained by the robust recursive least square method are more robust than those obtained by the standard recursive least square method .

该方法通过Huber函数选择加权因子，使得参数估计值对由测量系统引起的大扰动或异常值具有鲁棒性。

The method is with robustness for great disturbances or outliers generated by measurement systems via choosing weighting factor based on huber function .

多波束测深是一个动态的过程，采集得到的水深数据具有不可重复性，针对这一特点，将Bayes估计理论和MCMC抽样设计方法引入到测深异常值的探测中。

Multibeam echosounding is a dynamic process ; the depth is not repeating data . Aiming at the character ; the Bayes estimation theory and MCMC sampling design method are imported to detecting outliers .

该算法用Cauchy误差估计器代替传统的LMS误差估计器，提高了模型的鲁棒性，降低了训练样本异常值对收敛速度的影响。

In this algorithm , conventional LMS error estimator was replaced by Cauchy error estimator , the robustness of model was increased , and the influence of abnormal value in training samples on the convergence speed was reduced .

GSFC（9/80）地磁模型的评价和中国地磁台站异常值的确定

Evaluation of gsfc ( 9 / 80 ) model of geomagnetic field and determination of local anomaly solutions of observatories in China

该方法首先将时间序列映射到相空间，然后对相空间中的点实行一类分类，最后，根据KKT条件进行异常值检测。

The method first maps time series to phase space , and then obtains a distinguish function by one-class classification in phase space , finally , outliers detection is proceed based on KKT conditions .