Using Student Characteristics for Machine Learning Modeling in Higher Education Alumni Giving

Open Access
- Author:
- Mirabile, Dominic
- Area of Honors:
- Information Sciences and Technology
- Degree:
- Bachelor of Science
- Document Type:
- Thesis
- Thesis Supervisors:
- Vasant Gajanan Honavar, Thesis Supervisor
Andrea H Tapia, Thesis Honors Advisor
Andrea H Tapia, Faculty Reader - Keywords:
- machine learning
algorithms
higher education
fundraising
artificial intelligence - Abstract:
- In understanding complex relationships between variables and outcomes, the field of machine learning can be helpful for predicting and classifying new instances based on past data. This study applies a variety of machine learning algorithms to higher education alumni giving and compares the performance of each method in terms of classification accuracy and practicality for use. With increased reliance on private philanthropy, higher education must optimize their development efforts, which begins with the ability to identify likely donors. The models leverage the predictive power of student data to examine the link between student experience and alumni giving and to ensure the model could be applied to future classes using the standard set of collected data. The supervised machine learning methods employed in this study are: Naives Bayes, Decision Trees, Random Forest, Boosting, Bagging, and SMO Support Vector Machines. They represent diverse, widely used methods for classification and provide insight into the broader behavior of alumni giving. The results of the classifiers were largely consistent in terms of classification accuracy and area under ROC curve—suggesting that the modeling had reached the highest accuracy possible for this specific dataset. Still the machine learning implementations perform significantly better than chance and offer a valuable tool in focusing engagement efforts on likely donors. This study contributes to the related literature by offering a practical application of several types of machine learning in a growing field where data is available and rich. It supports the importance of collecting data about potential alumni donors while they are students and encourages the further integration of machine learning into prospecting techniques.