The test error is higher in predicting defaulted loan that is really paid rather than the opposite, and that is exactly what is better for the customer. The ROC (Receiver Operating Characteristic) confront FP Rate and TP Rate to establish the ability of the model to split the information correctly. Closer to 1 better. It is usually ok if the value is more than 0.7. Although in this case, the model is slightly under, the...
The test error is higher in predicting defaulted loan that is really paid rather than the opposite, and that is exactly what is better for the customer.
The ROC (Receiver Operating Characteristic) confront FP Rate and TP Rate to establish the ability of the model to split the information correctly. Closer to 1 better. It is usually ok if the value is more than 0.7. Although in this case, the model is slightly under, the complexity of the task can justify it.
Analyze and describe the level of accuracy the model achieves and the errors your model makes. Show a confusion matrix for each model. Are there any areas of the data where it performs worse than in others? Show a ROC curve for the decision as to whether or not a loan will be repaid and describe what the curve shows.
Ensemble methods that use a mixture of models, all built in different ways, to offer a panel of predictions rather than a single one. Using an ensemble introduces hyperparameters of its own such as how to combine the set of predictions into a single answer.
