Detecting Credit Card Frauds Using Isolation Forest And Local Outlier Factor - Analytical Insights

The usage of credit cards for online shopping and regular purchases is exponentially increasing and the fraud is related with it. A large number of fraudulent transactions are made every day. Moreover, it is become one of the rising issues. Therefore, it is basic that credit card organizations are skilled to perceive false transactions so clients are not charge for things which they didn't buy. Machine learning is one of the techniques considered as most successful to identify fraud. This paper offers proposed system for detection of fraud transactions with named FDS (Fraud Detection System) which can be used in a bank and in a organization where credit cards are used for transactions. Additionally, the paper illustrates various methods for fraud identification, compare all of them in terms of performance measures such as precision, and accuracy, to analyze the previous records of buyers and extract behavioral patterns. The authors have concentrated on the data pre-processing data sets just as sending of numerous anomaly identification algorithms such as Isolation Forest and Local Outlier Factor for calculation on the given PCA transformed exchange data that gives better result compare to usual classification algorithms.