Several machine learning models for detection of credit card fraud based on a transaction.
In this repository are several different machine learning models used to classify between fraud and non-fraud credit card transactions. Models such as linear regression, logistic regression, neural network, SVM, and K-means were implemented.
Outputs confusion matrix, per class accuracy, precision, recall, and error rates.
Several graphs can also be generated showing different error rates when changing different variables such as threshold of classification, learning rate, number of layers in neural network, etc.
Dataset used: https://www.kaggle.com/dalpozz/creditcardfraud
Labels were converted to numbers (0: non-fraud, 1: fraud) for easier processing.