Loan Approval from Historical Data

Background

Banks routinely lose money based on loans that eventually default. Per the Federal Reserve, at the height of the financial crisis in 2009-2010, the amount lost approached $500 billion. Most recently losses each quarter tend to approach $150 billion. Delinquency rates tend to be around 1.5% most recently. Because of this, it is vitally important for banks to ensure that they keep their delinquencies as low as possible.

• Can we accurately predict loan approval based on historical data?

• How can we confidentially determine whether a loan can be approved?

  • Rationale and Objective:
    • If a loan is current, the company is making profit and should approve such future loans based on the model.
    • If a loan is late or in default, the company is losing capital and should reject future loans based the model.

• What factors predict loan approval?

• Which variables best predict if a loan will be a loss and how much is the average loss?

Data

The data was retrieved from here. Lending Club is a peer to peer financial company. Essentially, people can request an unsecured loan between $1,000 and $40,000 while other individuals can visit the site to choose to invest in the loans. So, people are essentially lending to other people directly with Lending Club as a facilitator.

Methods

• Preprocess data in Python and R
• Variable selection

  • Python

    • SelectFromModel with XGBoost classifier utilizing GPU
    • VIF followed by Group Lasso

  • R

    • Model-Free Screening (MV-SIS)
    • Boruta
• Evaluate methods for class imbalance using:
  • Upsampling the minority class
  • Synthetic Minority Oversampling Technique (SMOTE)
• Test selected features using linear and non-linear ML algorithms
• Tune hyperparameters of different algorithms to increase predictive performance

Modeling

Machine Learning

Models were trained using the following libraries:

• XGBoost
• Catboost
• LightGBM
• RAPIDS: Logistic/Ridge/Elastic Net Regression, LinearSVC, Random Forest, XGBoost, K Nearest Neighbor
• SparkML: Logistic Regression, LinearSVC, Decision Trees, Random Forest, Gradient Boosted Trees
• Scikit-learn: Linear, Naive Bayes, Random Forest

Hyperparameter Optimization

For hyperparameter tuning, Optuna, Hyperopt, and GridSearchCV were utilized to explore the model parameters that resulted in the lowest error using various metrics for classification. Various trial/experiment sizes were completed to determine which parameters when incorporated into the model resulted in the lowest error.

Model Explanations

To explain the results from modeling, ELI5, SHAP and LIME were utilized.