Credit Risk Analysis

The purpose of this analysis was to use sampling algorithms and machine learning models to reduce bias and predict credit risk to help a Credit Card Lending company provide a more reliable loan experience and more accurate identification for loan candidates.

Project Overview:

1. Use Resampling Models to Predict Credit Risk
   • Naive Random Oversampling
   • SMOTE Oversampling
   • Undersampling
   • SMOTEEN (Over and Under Sampling)
2. Use Ensemble Models to Predict Credit Risk
   • Balanced Random Forest Classifier
   • Easy Ensemble AdaBoost Classifier
3. Interpret results and determine which supervised learning algorithm is best used for this dataset

Resources

• Source of data: LoanStats_2019Q1.csv
• Software: imbalanced-learn, skikit-learn, Python 3.7.10, Conda 4.10.3, Jupyter Notebook 6.3.0, Visual Studio Code 1.60.2
• Please see the refactored code here:
  • credit_risk_resampling.ipynb
  • credit_risk_ensemble.ipynb

Results

From our 6 different machine learning tests that when looking at balanced accuracy score, precision, and recall the Easy Ensemble Adaboost Classifier would be the best to choose at the following: - 93% Balanced Accuracy Score - 7% Precision - 91% Recall

Even though the Easy Ensemble Adaboost Classifier does not have a very reliable precision score out of the 6 machine learning models used the Easy Ensemble Adaboost Classifier would be the best model for credit card analysis within this dataset.