abduliante/vehicle-default-loan-prediction

Forecasting the likelihood of a customer defaulting their auto loan using classification models

vehicle-default-loan-prediction's Issues

Tuning parameters

Implement grid search on chosen models to choose best parameters:

• KNN
• Random forest
• XGBoost
• Logistics regression (Assigned to Mohammed)

Fix model deployment

Fix model deployment on heroku

Feature selection to choose best variables

Investigate logistics regression scores

Logistics regression with our chosen parameters yields these results:

Training classification report:

          precision    recall  f1-score   support

       0       0.65      0.49      0.56    135558
       1       0.63      0.77      0.69    153784

accuracy                           0.64    289342
macro avg      0.64      0.63      0.62    289342
weighted avg   0.64      0.64      0.63    289342

Validation classification report:

          precision    recall  f1-score   support

       0       0.81      0.40      0.54     34000
       1       0.24      0.67      0.35      9502

accuracy                           0.46     43502
macro avg      0.53      0.54      0.44     43502
weighted avg   0.69      0.46      0.50     43502

outliers

Remove outliers of the count of active loans and previously taken by the customer at the time of disbursement.