In this project, I will be using various machine learning models to predict credit risk on loan data. Due to the nature of the data's classification, it is often difficult to predict which loans are risky and which are good. I will be comparing the strengths and weaknesses of these models to determine how well they can perform in predicting credit risk.
AdaBoost Classifier
-
Accuracy score: 93%
-
Precision - for high risk: 0.09
-
Precision - for low risk: 1.00
-
Recall - for high risk: 0.92
-
Recall - for low risk: 0.94
-
The accuracy score is a measure of how well a model performs when it comes to making correct predictions. However, it is not enough to just see that the model gets it right. There must also be a total number of correct predictions.
-
The precision score is a measure of how well a model performs when it comes to identifying and validating positive classifications. A low precision means that a lot of false positives are generated.
-
The ability of a model to find all the positive samples is considered a high recall score. A low recall is an indication of a large number of false positives.
-
The true positive rate is a measure of how well a model performs when it comes to identifying and validating positive classifications. It is a weighted average of the precision score and the recall rate.