Using machine learning, this analyzes credit card risk using six machine learning models. We have employed different techniques to train and evaluate models with unbalanced classes. We are using imbalance-learn and scikit-learn libraries to build and evaluate models using resampling.
A low precision is indicative of a large number of false positives.
A low recall is indicative of a large number of false negatives.
| Model | Balanced Accuracy Score | Precision | Recall | F1 |
|---|---|---|---|---|
| Oversampling | 64% | 0.99 | 0.60 | 0.74 |
| SMOTE | 66% | 0.99 | 0.69 | 0.81 |
| ClusterCentroid | 66% | 0.99 | 0.40 | 0.56 |
| SMOTEENN | 54% | 0.99 | 0.57 | 0.72 |
| Balanced Random Forest | 68% | 1.00 | 1.00 | 1.00 |
| AdaBoost | 69% | 1.00 | 1.00 | 1.00 |
The models with the highest F1 score were Balanced Random Classifier and AdaBoost both with an F1 score of 1.00. However, SMOTE had an F1 score of 0.81 and a low recall of 0.69. Indicating that SMOTE is a more accurate model.
https://github.com/AJMnd/Credit_Risk_Analysis/blob/main/Resources/SMOTE.png
https://github.com/AJMnd/Credit_Risk_Analysis/blob/main/Resources/Rfc.png
https://github.com/AJMnd/Credit_Risk_Analysis/blob/main/Resources/Ada.png
