Isolation Forest and Xg Boosting for Classifying Credit Card Fraudulent Transactions
Chandra Sekhar Kolli1, T.Uma Dev2

1Chandra Sekhar Kolli, Research Scholar, Computer Science Department, GITAM Deemed to be University, Visakhapatnam, India and Assistant Professor, Koneru Lakshmaiah Education Foundation, Vaddeswaram, Guntur, India
2Dr. T. Uma Devi, Associate Professor, Computer Science Department, GITAM Deemed to be University, Visakhapatnam, India.
Manuscript received on 02 June 2019 | Revised Manuscript received on 10 June 2019 | Manuscript published on 30 June 2019 | PP: 41 -47 | Volume-8 Issue-8, June 2019 | Retrieval Number: E5795038519/19©BEIESP
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (

Abstract: Machine Learning based Neural Networks are extensively popular in creating machine learning models such as Text Classification, Face Recognition, Speech Recognition, Recommending new services etc. In the present hyper-competitive data-driven world these models are in huge demand since they provide high accuracy. In present days, the availability of internet and the wide variety of services like e-commerce, online shopping gained much popularity. On the other side of the coin, customers are facing adverse benefits due to fraudulent activities. Therefore, analyzing, detecting and preventing such unusual fraudulent activities in- real time is very essential and play crucial part. The main objective of this paper is to create a predictive model that capture the fraudulent transactions with high accuracy using Isolation forest & Local Outlier factor for detecting outliers that explicitly identifies anomalies and Extreme Gradient Boosting, an ensemble approach for constructing and evaluating the predictive model. A comparative study was done with existing models Logistic Regression, SVM, Random Forest with Extreme Gradient (XG) Boosting algorithm. The proposed model XG Boosting shows better performance and secured high accuracy 0.98.
Keyword: Fraudulent Activities, Isolation Forest, Local Outlier Factor, Extreme Gradient Boosting, ROC.
Scope of the Article: Classification