Credit_Risk_Analysis

Analysis Overview

The purpose of the analysis is to use different machine learning methods to find the best algorythm to use for predicting credit risk

• Results

1. The first method is the naive oversampling method Balanced accuracy score for this method is 65.15% with a precision of .99 and a recall of .68
   
2. the 2nd method is the SMOTE Oversampling method with a balanced accuracy score of .624 and a precision of .99 and a recall of .66
   
3. the 3rd method is the undersampling method with a balanced accuracy score of .624 and a precision of .99 and a recall of .44
   
4. The 4th method is the combination over and under sampling method with a balanced accuracy score of .64 and precision of .99 and a recall of .58
   
5. the 5th method is the Random Forest Classifier with a balanced accuracy score of .7877 and a precision .99 and a recall of .91
   
6. the 6th method is the EasyEnsemble Ada Boost Classifier with a balanced accuracy score of .925 and a precisison of .99 and a recall of .94
   

Summary

In summary after sampling all the methods I would recommend the Easy Ensemble method because its accuracy and recall scores are so much higher than all the other methods