Welcome to the exciting world of credit card fraud detection! This project uses PySpark and Google's Dataproc to build a model that can a ccurately identify fraudulent transactions and prevent them from happening.
Our dataset comes from Kaggle and consists of over 285,000 credit card transactions, out of which only 492 are fraudulent. The data has been anonymized to protect privacy, but we can still use it to train our model and detect fraud with impressive accuracy.
To run this project, you'll need to have PySpark, Jupyter Notebook, and Google's Dataproc installed. If you don't already have them, don't worry! It is easy to put all in motion.
We're happy to report that our final model achieved an F1 score of 0.82, with an accuracy of 99.93%. That means we were able to detect 73% of all fraudulent transactions, with a false positive rate of only 0.01%.
In conclusion, this project showcases the power of PySpark, Google's Dataproc, and machine learning in the fight against credit card fraud. It was a challenging task, with an imbalanced dataset and a high-stakes goal, but we rose to the challenge and built a model that can make a real-world impact.
So, what are you waiting for? Join us on this exciting journey of fraud detection and prevention, powered by Google's Dataproc!