Class Outlier Mining:
Nabil M. Hewahi and Motaz K. Saad
Motaz K. Saad
International Journal of Intelligent Technology, Vol. 2, No. 1, pp 55-68, 2007
• In large datasets, identifying exception or rare
cases with respect to a group of similar cases
is to be considered very significant problem.
• The traditional problem (Outlier Mining) is to
find exception or rare cases in a dataset
irrespective of the class label of these cases,
they are considered rare event with respect to
the whole dataset.
• Present an overview of Class Outlier.
• Introduce a novel definition of a class outlier
and propose COF factor.
• Propose a new algorithm for class outlier
• Present experimental results.
• Perform a comparison study.
• An Outlier is a data object that does not
comply with the general behavior of the
data (unusual pattern)
• It can be considered as noise or
exception but is quite useful in fraud
detection and rare events analysis.
• It is the problem of detecting rare
events, deviant objects, and
• Is an important data mining issue in
knowledge discovery; it has attracted
increasing interests in recent years.
Outlier Mining: Business Applications
• Fraud detection
• Credit approving
• Stock market analysis
Identifying computer network intrusions
• Data cleaning
• Surveillance and auditing
• Health monitoring systems
Insurance, banking, money laundering
telecommunication ..., etc).
Outlier Detection Methods
• Statistical based (Distribution based)
– K Nearest Neighbors (KNN)
• Model‐Based (Neural Network): Replicator
Neural Network RNN
Statistical (Distribution) based
Outlier Detection Method
Variation of Distance‐Based Approach
for detecting Outlier
NN Model for Detection Outlier
A schematic view of a fully connected R