Building a ML Model to Predict Student Enrollment at The School of the Art Institute

In honor of SPR’s 50^th anniversary, we’re donating $50,000 worth of machine learning (ML) services to The School of the Art Institute of Chicago (SAIC)! When we created this 50K giveaway, our goal was to partner with a local organization that thinks implementing an ML-based solution could make a significant, positive impact on their organization and the community – and we believe SAIC is doing just that.

The School of the Art Institute of Chicago is a prestigious institution renowned for its commitment to fostering artistic talent. Like many educational institutions, SAIC faces the challenge of optimally allocating its limited resources. The school’s primary objective in applying machine learning technology is to improve its capacity to predict the decisions of students who are offered admission. They want to know which students accepted will ultimately enroll for the ensuing academic term. By doing so, SAIC can better manage its resources and maintain its dedication to accepting the most talented, promising, and diverse students.

To answer this question, we’re gathering and examining a myriad of data points throughout the admissions process. Key data sources include:

• The Common Application: A widely used platform that collects standardized information about prospective students.
• The candidate management system: Details of the application process.
• Students’ artistic portfolios: These portfolios, integral to the admissions process at SAIC, offer rich insights into students’ creativity, skill, and dedication to the arts. By incorporating portfolio data into the analysis, SPR and SAIC can form a more nuanced understanding of students’ profiles and potential.
• Student interactions with the university such as participation in on-campus events: A student’s decision to enroll may be influenced by their experiences with and perceptions of the university environment.

By examining these data points, the project aims to uncover potential correlations between student engagement and enrollment decisions.

Once collected, the data is then explored to determine what relationships exist between a student and their propensity to enroll. From there the data is processed to be most effectively used by the various algorithms. When the initial models are trained, the results are examined to understand the performance of the model and subsequent models are optimized based upon the initial models results. This project shows how machine learning can help when making strategic decisions. There is value in a holistic approach – one that incorporates many data points, from standardized application information to specific student-university interactions.

This project will boost SAIC’s admissions strategy and, because we’ll be implementing a binary classification-type model, the overall approach and technique we are planning can be adapted and applied effectively to just about any business that wants to classify or categorize data. A popular implementation in business is to classify customers that are likely to churn (fancy data science language for customers that will leave).