Edgewise outliers of network indexed signals
Preprint
Abstract
We consider models for network indexed multivariate data involving a dependence between variables as well as across graph nodes. In the framework of these models, we focus on outliers detection and introduce the concept of edgewise outliers. For this purpose, we first derive the distribution of some sums of squares, in particular squared Mahalanobis distances that can be used to fix detection rules and thresholds for outlier detection. We then propose a robust version of the deterministic MCD algorithm that we call edgewise MCD. An application on simulated data shows the interest of taking the dependence structure into account. We also illustrate the utility of the proposed method with a real data set.Keywords
All Related Versions
This publication has 22 references indexed in Scilit:
- Minimum covariance determinant and extensionsWIREs Computational Statistics, 2017
- Comparison of local outlier detection techniques in spatial multivariate dataData Mining and Knowledge Discovery, 2016
- Spatial sign correlationJournal of Multivariate Analysis, 2015
- Multivariate Spatial Outlier Detection Using Robust Geographically Weighted MethodsMathematical Geosciences, 2013
- Identification of local multivariate outliersStatistical Papers, 2013
- An Overview of Approaches to the Analysis and Modelling of Multivariate Geostatistical DataMathematical Geosciences, 2011
- Minimum covariance determinantWIREs Computational Statistics, 2009
- On Detecting Spatial OutliersGeoInformatica, 2007
- Random Graphs for Statistical Pattern Recognition2004
- Multivariate Estimation with High Breakdown Point1985