University's Donor - Predictive Analytics

Summary

This project solely focuses on predicting which individuals related to the University - students, alumni, staffs, former staffs, etc - that will most likely donate their assets and/or money to the University.

The main challenge of this project is that there is only a handful of the observations that has donated to the university, in other words there is a huge imbalance in the proportion of the target variables. Less than 1% of the total observations have donated to the University and more than 99% have not.

Synthetic Minority Oversmapling Technique (SMOTE) is used to balance the proportion of the target variable before training various machine learning algorithms to create predictive models.

R Codes on GitHub