Rank-based outlier detection

Abstract

ABSTRACT:We propose a new approach for outlier detection, based on a new ranking measure that focuses on the question of whether a point is "important" for its nearest neighbors; using our notations low cumulative rank implies the point is central. For instance, a point centrally located in a cluster has relatively low cumulative sum of ranks because it is among the nearest neighbors of its own nearest neighbors. But a point at the periphery of a cluster has high cumulative sum of ranks because its nearest neighbors are closer to the points. Use of ranks eliminates the problem of density calculation in the neighborhood of the point and this improves performance. Our method performs better than several density-based methods, on some synthetic data sets as well as on some real data sets. KEYWORDS: AbstractWe propose a new approach for outlier detection, based on a new ranking measure that focuses on the question of whether a point is "important" for its nearest neighbors; using our notations low cumulative rank implies the point is central. For instance, a point centrally located in a cluster has relatively low cummulative sum of ranks because it is among the nearest neighbors of its own nearest neighbors. But a point at the periphery of a cluster has high cummulative sum of ranks because its nearest neighbors are closer to the points. Use of ranks eliminates the problem of density calculation in the neighborhood of the point and this improves performance. Our method performs better than several density-based methods, on some synthetic data sets as well as on some real data sets.

A Comparison of Outlier Detection Techniques for High-Dimensional Data

“…As an example, Huang et al 10 developed an outlier detection algorithm called RBDA (Rank Based Detecting Algorithm). It takes the ranks of each object in its neighbors as the proximity degree of the object.…”

Association Measures in Network Outlier Detection Methods

“…The main objective of detecting outliers is to retrieve the objects from the large datasets that have different behavior than the normal object present in the data [8]. Detection of outliers is the important field of data mining for which various algorithms have been used [9].In literature, various outlier detection algorithms are used [10]. Outlier detection has been studied on a large variety of data types including high-dimensional data, uncertain data, stream data, graph data, time series data, spatial data, and spatio-temporal data.…”

Ranking with Distance based Outlier Detection Techniques: A Survey

“…In this approach author proposed that each data point of the given data set should be assigned a degree of outlines and they refer it as the "Local Outlier Factor" (LOF) [20,21] of the data point and it is calculated as given below.…”

“…is calculated for selected values of k in a pre-specified range, max L k (p) is retained , and a p with large LOF is declared to be outlier [20] 3…”