A Multiple Classifiers System for Anomaly Detection in Credit Card Data With Unbalanced and Overlapped Classes

Abstract

Frauds and default payments are two major anomalies in credit card transactions. Researchers have been vigorously finding solutions to tackle them and one of the solutions is to use data mining approaches. However, the collected credit card data can be quite a challenge for researchers. This is because of the data characteristics that contain: (i) unbalanced class distribution, and (ii) overlapping of class samples. Both characteristics generally cause low detection rates for the anomalies that are minorities in the data. On top of that, the weakness of general learning algorithms contributes to the difficulties of classifying the anomalies as the algorithms generally bias towards the majority class samples. In this study, we used a Multiple Classifiers System (MCS) on these two data sets: (i) credit card frauds (CCF), and (ii) credit card default payments (CCDP). The MCS employs a sequential decision combination strategy to produce accurate anomaly detection. Our empirical studies show that the MCS outperforms the existing research, particularly in detecting the anomalies that are minorities in these two credit card data sets. INDEX TERMS Anomaly detection, credit card, multiple classifiers.

Show Me Your Claims and I'll Tell You Your Offenses: Machine Learning-Based Decision Support for Fraud Detection on Medical Claim Data

“…We analyze the performance of our fraud detection approach by applying the developed model to 1,849 preprocessed and unseen records. Therefore, the Area Under the Receiver Operator Curve (AUC), Precision, and Recall are used as evaluation metrics since they are well known and applied in multiple related publications (e.g., Kalid et al [39] and Bauder et al [1]). In addition, this enables a comparative assessment against previously developed and upcoming approaches.…”

“…Based on this, AUC considers Recall and Precision, thus being more balanced than Accuracy [16]. It is defined as the area under the Receiver Operator Curve (ROC), representing the capability of a classifier in differentiating classes [39]. The ROC is created by plotting Recall against Precision.…”

Minimizing the Overlapping Degree to Improve Class-Imbalanced Learning Under Sparse Feature Selection: Application to Fraud Detection

“…Other studies proved that the influence of imbalance alone on the performance of classifiers is limited. However, the overlap degree influences the performance of classifiers during training [9], [10]. In the same context, to measure the overlap degree, a new metric called the augmented R-value has been proposed.…”

A New Residual Dense Network for Dance Action Recognition From Heterogeneous View Perception

“…Zhang and Zhang ( 2011 ) proposed a selective ensemble learning algorithm to improve the prediction and classification efficiency of the ensemble learning machine and reduce its storage requirements. Kalid et al ( 2020 ), the generalized combination rule of multi-classifier system based on genetic algorithm for parameter estimation was adopted to carry out pattern classification. Heng et al ( 2016 ), the output of multiple extreme speed learning machines was used for simple mean algorithm fusion processing, and the recognition accuracy of the final model output was 3.6% higher than that of a single extreme speed learning machine.…”