Optimizing Fraud Detection In Financial Transactions: A Comprehensive Exploration Of The Effectiveness Of Random Forest And Isolation Forest Algorithms In Detecting Anomalies Within Credit Card Transactions
Main Article Content
Abstract
Identifying fraudulent activity during credit card transactions is vital to safeguarding the efficiency of financial systems. Using credit card transaction data, we examine and analyze the performance of two common anomaly detection methods: Random Forest and Isolation Forest. Isolation Forest and Random Forest are the names of these algorithms, respectively. The approaches for gathering datasets, training models, and testing performance using multiple metrics are all discussed in the experimental procedures. The results suggest the efficacy of both strategies in spotting fraudulent transactions, with each strategy providing a distinct set of qualities that set it apart from the others. Through a rigorous review of performance data, we share insights into the capabilities of each strategy in this paper. The F1-score, recall, accuracy, precision, and Matthew’s correlation coefficient are some of these metrics. We also examine the repercussions of the data, suggest vital research problems, and give advances to fraud detection systems. The study examines each of these difficulties.