Open AccessMultilevel Fraud Detection System using Voting Techniques
Author(s) -
Anjali Agrawal*,
Mahesh Parmar
Publication year - 2019
Publication title -
international journal of innovative technology and exploring engineering
Language(s) - English
Resource type - Journals
ISSN - 2278-3075
DOI - 10.35940/ijitee.l2840.1081219
Subject(s) - computer science , random forest , feature selection , data mining , naive bayes classifier , anomaly detection , voting , weighted voting , data set , feature (linguistics) , set (abstract data type) , pattern recognition (psychology) , artificial intelligence , support vector machine , politics , political science , law , programming language , linguistics , philosophy
Fraud detection is an enduring topic that pose a threat to the information security systems. Data mining helps detect and rapidly identify fraud and take instant action to reduce losses. It analysis the various forms of attacks through classifications algorithms of data mining. In this dissertation the data set of NSL-KDD is examined in identifying anomalies in network traffic patterns and the importance of various classification techniques is studied. Firstly the data is classify in two classes Normal and Anomaly and after then the anomaly data categories in various Attack forms i.e Probe, U2R, R2L, DOS to detect the fraud. The analysis performed through feature selection and classification techniques present in WEKA tool of data mining. The features are extracted through feature selection (CfsSubsetEval, InfoGain, FilterSubsetEval and FilterAttributeEval) methods from 41 attributes to 11 attributes using CfsSubsetEval with best first Search and after extraction the features the classification algorithms(Naïve Bayes, RC, RT, RF and DT) applied for finding Accuracy. The best Result in terms of Accuracy is finding through the voting method using Random Committee and Random Forest is 99.94%.