In this project, I work on the problem of Credit Card Fraud Detection. The data is highly imbalanced with the positive class (fraud) accounting merely for 0.172% of the data. In classification problem balancing your data is extremely important. Here I explore how accuracy should not be the only criteria to judge model performance.
You can get the dataset from Kaggle: https://www.kaggle.com/mlg-ulb/creditcardfraud/downloads/creditcardfraud.zip/
This notebook was in support of my article on Balancing data using SMOTE. Find it at https://medium.com/analytics-vidhya/balance-your-data-using-smote-98e4d79fcddb