My steps:
- Import the modules
- Read the Dataset
- Check for relative proportions of Fraudulant Cases
- Plot the named features
- Reorder the columns Amount, Time then the rest
- Plot the distributions of the features
- Check for null values
- Separate response and features (Undersampling before cross validation leads to overfitting)
- Use SkLearn for splitting
- Create the cross validation framework
- Import the imbalance learn module, classifiers, metrics
- Import Random Forest
- Used metrics such as accuracy, precision, recall, F1 score
- Define get_model_best_estimator_and_metrics
- Cumulatively create a table for the ROC curve