Credit risk is an inherently unbalanced classification problem, as good loans easily outnumber risky loans.
Using a credit card credit dataset and Python, several machine learning modules will be used to evaluate and predict credit risk.
Techniques used to predit credit risk:
- Oversample the data using the RandomOverSampler and SMOTE algorithms.
- Undersample the data using the ClusterCentroids algorithm.
- A combinatorial approach of over- and undersampling using the SMOTEENN algorithm.
- Compare two new machine learning models that reduce bias, BalancedRandomForestClassifier and EasyEnsembleClassifier, to predict credit risk.
Once these models have been completed, their performance will be evaluated and a written recommendation will be made on whether they should be used to predict credit risk
- Deliverable 1: Use Resampling Models to Predict Credit Risk
- Deliverable 2: Use the SMOTEENN Algorithm to Predict Credit Risk
- Deliverable 3: Use Ensemble Classifiers to Predict Credit Risk
- Deliverable 4: A Written Report on the Credit Risk Analysis
Data Sources: - LoanStats_2019Q1.csv
Software: - Jupyter Notebook 6.1.4 - Python 3.8.5
- Accuracy Score for the RandomOverSampler model is 63%
- The precision for the high-risk is 1% and F1 score is 2%, which are not good enough to state that the model will be good at classifying.
- The accuracy score of the SMOTE model is a little bit better than the RandomOverSampler.
- The precision for the high-risk is very low at 1%, indicating a large number of false positives, which indicates an unreliable classification.
- The F1 score is 2% which also very low.
Overview of the analysis: Explain the purpose of this analysis.
Results: Using bulleted lists, describe the balanced accuracy scores and the precision and recall scores of all six machine learning models. Use screenshots of your outputs to support your results
Summary: Summarize the results of the machine learning models, and include a recommendation on the model to use, if any. If you do not recommend any of the models, justify your reasoning.