Comparative Analysis of Point Outlier Detection using Cluster Based Techniques
Date
2019
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Department of Computer Science & Information Technology
Abstract
Outlier detection is the process of finding peculiar pattern from given set of data. Nowadays,outlier detection is more popular subject in different knowledge domain. Data size is rapidly
increases every year there is need to detect outlier in large dataset as early as possible.
In this research, comparison of three different cluster based outlier detection algorithm i.e. Kmeans
with
OD,
Partition
around
medoids
(PAM)
with
OD
and
density
based
spatial
clustering
algorithm
with
noise
(DBSCAN)
is
presented.
The
main
aim
of
this
research
is
to
evaluate
their
performance
of those three different cluster based outlier algorithm for different dataset with
different dimension. The dataset used for this research are chosen such way that they are different
in size, mainly in terms of number of instances and attributes. When comparing the performance
of all three cluster based outlier detection algorithms, in average PAM with OD is found to bebetter algorithm to detect outlier in most cases with accuracy level 98.30% as well as 92.20%
precision, 95.30% recall and 93.43% F-measure value.
Description
Keywords
Outlier detection, K-means with OD, PAM with OD, DBSCAN (Density based spatial clustering algorithm with noise)