A COMPARATIVE STUDY: CREDIT CARD FRAUD DETECTION USING MACHINE LEARNING

For both online and offline purchases, Credit cards are one of the most used and popular modes of payment which is also leading to increasing daily fraud transactions.It results in huge financial losses as these events take place frequently. Number of online transactions has grown in large quantities where online transactions hold a huge share of these transactions. Thus, credit card fraud detection applications are offered by banks and financial institutions. In order to maintain the reliability of the payment system, an efficient fraud detection methodology is essential. Various Machine learning algorithms have increased the capacity of handling a huge dataset which helps in detecting many fraudulent transactions very efficiently along with its high power of processing and computing data. For real time problems which occur in our daily life Machine Learning provides fast and efficient solutions. In this comparative study, credit card fraud detection is performed on the huge datasets which involves lakhs of rows of transaction data by using various Machine Learning Algorithms as it consumes lot of time to manually check and find the patterns in the data for fraudulent transactions. Due to an imbalanced dataset, SMOTE technique is used for oversampling. Further, feature selection is performed and the dataset is split into two parts, training data and test data. Machine Learning models used are: Logistic Regression (LR), Naive Bayes (NB), K-Nearest Neighbors (KNN), Random Forest (RF), Artificial Neural Networks (ANN).