APPLICATION
OF DATA MINING : FRAUD DETECTION

Abstract

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!


order now

The
application of Data Mining techniques
for the detection of Financial Fraud:
The
Data Mining techniques in the field of Fraud Detection was analyzed and categorized into
four categories of financial fraud (securities and commodities fraud, insurance fraud, bank fraud, and
other related financial fraud) and six classes of data mining techniques (regression, classification, clustering, outlier detection, prediction, and
visualization). The
Data Mining approach have been applied wide-ranging to the detection of credit
card fraud, although insurance fraud and cooperate fraud have also attracted a
skillful trade of attention in recent years. In difference, we find a distinct
lack of research on money laundering, securities and commodities fraud  and mortgage fraud. The major data mining
techniques used for FFD are neural networks, logistic models, decision trees,
and the Bayesian belief network, all of which them provide key solutions to the
problems inherited in the detection and in the classification of fraudulent
data.

Introduction

“The Association
of Certified Fraud Examiners (ACFE) defined fraud as: The use of one’s
profession for personal enrichment through the deliberate misuse or application
of the employing organization’s resources or assets.” In the technological
systems, fraudulent activities have occurred in many areas of daily life such
as mobile communications, telecommunication net works, E-commerce and online
banking. Fraud detection includes identification of fraud as fast as possible
once it has been accomplished. Fraud detection methods are continuously
designed to defend criminals in reinventing to their strategies. The development
of new fraud detection methods is made more difficult due to the severe
limitation of the exchange of ideas in fraud detection. At current, fraud detection
has been applied by a number of methodology such as artificial intelligence, statistics
and data mining.

 

 

 

 

Data Mining Techniques
In Fraud Detection

 

1.   
Credit Card Fraud Detection:

The
Credit Card Fraud Detection Challenge involves replication of  past credit card transactions with the
expertise knowledge of the ones that turned out to be fraud. This replica is
then used to distinguish whether a new transaction is fraudulent or not. Our
goal here is to determine 100% of the fraudulent transactions while minimizing
the inaccurate fraud classifications.

Credit card fraud detection is fully
confidential and is not much reveal in open. Large span data-mining techniques
can enhance the state of the art in commercial exercise. Scalable techniques to
examine heavy amounts of transaction data that effectively calculate fraud
detectors in a timely manner is a major problem, specifically for e-commerce.
Besides efficiency  and scalability , the
fraud-detection task presents technical problems that include skewed division
of training data and non uniform cost per error, both of which have not been
widely studied in the knowledge-discovery and data mining network. Following
are the techniques that detect credit card fraud detection-

1.1.        
Outlier
Detection

1.2.        
Neural
Networks

 

2.   
Computer Intrusion Detection:

An intrusion detection system is
required to perform and automate system  observing
by keeping aggregate survey trail statistics. Intrusion detection techniques
can be widely categorized into two categories depends on model of intrusions:
misuse and anomaly detection.

Misuse Detection- Misuse
detection trials to recognize the attacks of earlier observed intrusions in the
form of a signature or a pattern (for example, reoccurring changes of directory
or attempts to read a password file) and forthrightly monitor for the
occurrence of these patterns. Misuse approaches include model-based reasoning, keystroke
dynamics monitoring, expert systems, and state transition analysis. Misuse
detection is simple than Anomaly detection.

Anomaly detection- Anomaly
detection approaches include predictive pattern generation, neural net- works,
and statistical approaches. The advantage of anomaly detection is that it is possible
to find out novel attacks against systems.            

Following are the techniques that
detects computer intrusion detection-

2.1.        
Expert
Systems

2.2.        
Neural
Networks

2.3.        
Model
Based Reasoning

 

3.   
Telecommunication Fraud Detection

Large
amounts of data are being collected as a result of the large usage of mobile
telecommunications. Over a couple of time, an individual person phone generates
a largish pattern or signature of use. While call data are measured for
individual subscribers about the data indicative of fraudulent call signatures
or patterns. Furthermore, examining is thus needed to be able to isolate
fraudulent use. An unsupervised learning algorithm can examine and cluster call
signatures or patterns for an individual subscriber in order to simplify the
fraud detection process. This research investigates the unsupervised learning
potentials of two neural networks for the profiling of calls made by users over
a couple of time in a mobile telecommunication network. Our study provides a
comparative examination and application of Long Short-Term Memory (LSTM) and
Self-Organizing Maps (SOM) recurrent neural networks algorithms to user call
data records in order to obtain a descriptive data mining on users call patterns.

Following
are the techniques that detects telecommunication fraud detection-

3.1.        
Rule-Based
Approach

3.2.        
Neural
Network

3.3.        
Visualization
Methods