Analyzing the Classification Accuracy of Deep Learning and Machine Learning for Credit Card Fraud Detection
Mohammad Naveed Hossain, Md. Mahedi Hassan, Raiyan Janik Monir · ASIAN JOURNAL OF CONVERGENCE IN TECHNOLOGY · 2022
The purpose of this study is to classify a dataset of credit card security problems by employing six different machine learning (ML) approaches. The Support Vector Machine (SVM), Random Forest (RF), Bagged Tree, K-Nearest Neighbor (KNN), Naive Biased Classifier, and Extreme Gradient Boosting were selected as the classifiers to use (XGBoost). The classification accuracy of the machine learning algorithms was compared with that of a technique for categorization that is based on deep learning called Long Short-Term Memory (LSTM). The KNN machine learning approach had a maximum accuracy of 97.50 percent, while the LSTM machine learning method had an accuracy of more than 96 percent and promised to give biologically appropriate control of upper-limb movement. In addition to enhancing accuracy, the research has investigated how the effects of removing the channel with the most noise from the algorithms can have on accuracy. This was done in an effort to handle data in a more effective manner.