Auto Loan Classificataion

A machine learning project with 5 other graduate students and mentored by a senior data scientist from Colombia Threadneedle

Background Information

Hi, Welcome to my Github page - Auto Loan Classification Prediction. As a team, we are predicting an auto loan default classifier with 1 as default and 0 as no default. We are interested also in the profile differences between prime loan borrowers and subprime loan borrowers.

Language Used

• Jupyter Notebook: Python

Computing Machine Used

• Google Cloud Platform Virtual Machine (8 vCPUs, 32 GB memory)
• apply tmux to prevent VM from crashing

Machine Learning Methods Used

• Decision Tree
• Random Forest
• XGboost
• Logistics Regression
• Naive Bayes
• KNN

Files Desriptions

All databases are stoed here

• autodrive20171_database.csv: Santander (subprime auto) DB scraped and aggregated for ML purposes.
• database_gmal1801.csv: General Motor (prime auto) DB scraped and aggregated for ML purposes.
• cti_ml_final.ipynb: Python code in Jupyter environment used to run classification prediction for all 6 models mentioned above.