Destructive Outcomes of Digitalization (Credit Card), a Machine Learning Analysis

Abstract

AbstractAs the phase of digitization reaches our day-to-day lives, things are easily available and accessible by the computer, which is quite easier and faster method for transactions. The pandemic also played a huge role in growth in credit card fraud activities. And, that has led to massive increase in credit card fraud dramatically. As a result, fraud detection should include surveillance of the person’s/spending customer’s attitude in order to determine, prevent, and detect unwanted behavior. For both online and in-person buying, credit cards are the most convenient way of payment. Fraud detection is agitated not only with capturing fraudulent activities, but also with discovering them as early as possible, since this kind of fraud costs millions of dollars of people. Machine learning algorithms have proven to be extremely useful in detecting fraud of smart cards. Because of the uneven nature of regular classification algorithms, they are ineffective in detecting credit card fraud. Isolation forest algorithm is been used in the proposed scheme, and the local outlier (Tripathi et al. in J Pure Appl Math 118:229–234, [1]) factor is used to recognize fraudulent transactions and their accuracy.KeywordsMachine learningImbalanced dataSmart cardFraud detectionIsolation forest modelLocal outlier factor