Kaggle Credict Card Fraud

Topics Covered

• Handling imbalanced data
• Undersampling and Oversampling data
• SMOTE for oversampling
• Analysis using TSNE
• SVM Classifier
• KNeighborsClassifier Classifier
• Neural Network for classification
• Auto Encoders for Anomoly Detection
• Metrics: ROC, Precision, Recall

How to run

1. Build docker image

   docker build -t creditcardfraud .

2. Run docker container

   docker run -it -v $(pwd):/workdir -p 8888:8888 -t creditcardfraud bash

3. From inside the container workdir run to start jupyter notebooks

   jupyter-notebook --ip='0.0.0.0' --port=8888 --no-browser --allow-root

Resources

• Dataset