Recently, there has been an increase in the number of transactions involving digital money, which has been accompanied by an increase in the total number of transactions. Learning by machine is dependent on the data that is provided to it. Credit card fraud statistics is biased, according to the data. Due to the limited number of fraudulent transactions, a skewed data set produces findings that are not optimal for the various machine learning algorithms under consideration. Using the outcomes of under-sampling and over-sampling, a comparison is made between the various algorithms that are taken into consideration. Random Forest, Adaboost classifier, Stochastic Gradient Descent classifier, K Nearest Neighbour, Logistic Regression, Gaussian Naive Bayes, Decision Tree, and Support Vector Machine are some of the methods that have been investigated. Distinct results are obtained by under-sampling and over-sampling, respectively. When it comes to undersampling, the Adaboost classifier produces positive results, whereas the Random Forest method exhibits remarkable performance when it comes to oversampling. There are a number of metrics for performance that are used to evaluate algorithms.