Machine Learning For Credit Card Fraud Detection System

The quick growth of the e-commerce industry led to an exponential growth in credit card online purchases, resulting in an increase in fraud. Banks have become increasingly important in recent years. Fraud in the credit card system is difficult to identify. In order to detect credit card fraud, machine learning is used for the business transactions. Banks utilise various machine learning approaches to forecast these transactions; historical data has been collected, and new characteristics have been added to improve accuracy. The ability to predict the detection of fraud performance in the sampling has a significant impact on credit card transactions. Using this method we can data-set, variable selection, and detection utilised techniques. To balance the data set, we have used oversampling, which resulted in 60% of fraudulent transactions and 40% of legitimate transactions. These three are techniques that are used for the dataset, and to complete the work. Implementation is done using the R programming language. The effectiveness of the methods is evaluated on the basis of sensitivity, specificity, precision and error rate for different variables. The logistic regression, decision tree and random forestry classifications are respectively 90.0, 94.3 and 95.5. The Random forest outperforms the logistic regression and decision tree procedures, according to the results.