Detecting fraudulent credit card transactions using Machine Learning algorithms.
[Dataset] https://www.kaggle.com/mlg-ulb/creditcardfraud
[Summary] This dataset contains 492 fraudulent, 284,315 normal and 284,807 transactions in total, made by european credit card holders in 09/2013. Since the two classes (fraud and normal transactions) are unbalanced, we use synthetic minority over-sampling technique (SMOTE) to even out the two classes. We fit this balanced dataset to six different machine learning models (Logistic regression, Naive Bayes, Decision tree, Random forest, Neural network (MLP classifier) and XGBoost) and compare their performance using the area under the ROC curve (AUC).
[ML Algorithms]
- Logistic regression
- Naive Bayes
- Decision tree
- Random forest
- Neural network (MLP classifier)
- XGBoost