Predictive modeling for alumni donations
LE3 .A278 2016
Bachelor of Science
Mathematics and Statistics
Mathematics & Statistics
Donations made to Acadia University are an important factor in helping students succeed through scholarships and research, as well as in preserving facilities. Aca- dia University's Advancement Office keeps records of these donations and the in- formation of donors over time. A large portion of donors consist of Acadia alumni who are giving back to the university. An objective for the Advancement Office is to encourage the alumni who donated in the past to continue donating, and to recruit new alumni donors through donation campaigns. Since there are many alumni on record, it is too costly to contact every one of them. By using data on alumni provided by the Office of Advancement, it is possible to highlight a group of likely donors to target through donation campaigns. The data includes information on the alumni themselves (such as gender, location and profession), the degrees that they have obtained at Acadia University (type of degree, faculty and graduation year), as well as information on the donations already made. This analysis will implement a random forest algorithm to predict how likely an alum- nus is to donate in the next 5 years, and generalized additive models to predict the amount an alumnus would donate. The analysis further explores how alum- nus information, degree history and donation history affect the future donation probability and the predicted amount.
The author retains copyright in this thesis. Any substantial copying or any other actions that exceed fair dealing or other exceptions in the Copyright Act require the permission of the author.