This is our final year B.tech project at IET Lucknow.
The datasets contains transactions made by credit cards in September 2013 by European cardholders. This dataset presents transactions that occurred in two days, where we have 492 frauds out of 284,807 transactions. The dataset is highly unbalanced, the positive class (frauds) account for 0.172% of all transactions.
We have used a total of 5 algorithms in our project
- Logistic Regression
- K-Nearest Neighbours
- Decision Tree
- Random Forest
- XgBoost
6 Techniques are used for undersampling or oversampling
- Random oversampling
- SMOTE Oversampling
- Random Undersampling
- Tomek Links Undersampling
- Cluster centroids undersampling
- SMOTE + Tomek Links
We have compared each of the 5 algorithms in all 7 scenarios comparing their accuracy through Area under curve of ROC Curve.