Credit Card Fraud Classification

In this project, we attempt to identify fraudulent credit card transactions. We approach the problem as a heavily imbalanced classification task.

We use the following resampling approaches:

• random undersampling
• random oversampling
• SMOTE
• ADASYN

Additionally, we use deepchecks to investigate how the resampling impacts the distribution of the features in the training data.

For more voice-over, please refer to the following article.

References

• data: https://www.kaggle.com/datasets/mlg-ulb/creditcardfraud