New Predictive Modeling Aims to Improve Forecasting and Decision-Making in Higher Education

By Alan Mlynek

University partners using Noodle Manage (N. Manage) can now take advantage of a nascent partnership between Noodle and a new wave of AI-driven data modeling firms to quickly create data models based on the university data already aggregated and organized through N. Manage. These models can answer a range of questions that are critical to effective university planning and leadership.

Questions to Answer

Aggregates for Operations and Planning

• How do universities use all available data to dynamically forecast new enrollments, re-enrollments, and the total student census?
• How should forecasts inform revenue projections, course section demand, and university staffing?

Student-level Analysis for Intervention

• How do you identify the prospective students that are most likely to enroll if they receive personal outreach from a university? How do you use prospect behaviors to tailor a message for that outreach?
• How do you identify the existing students that may or may not re-enroll who would most benefit from personal outreach from a university? How do you use student data to identify the impediments to re-enrollment and tailor support to overcome obstacles?

Partnership with Noodle

Universities engaging Noodle for predictive modeling can expect the following during our engagement. Noodle will deliver:

1. Preliminary scoping to fine-tune the question the model is meant to answer and the way the university will apply the results to their operations
2. Data engineering, as necessary, to organize the necessary university data through Noodle’s proprietary data lake and schema to feed the modeling engine the correct information
3. Configuration of the data modeling tools to produce multiple models that address the underlying research question
4. Recommendation of a data model for the university to operationalize with a review and analysis of the modeling results
5. Provision of new imputed fields (e.g., enrollment propensity score) to operational systems (e.g., Slate)

How is this Different from ‘Big Data’ Exercise of Yesteryear?

Application of AI tools in the data science space has reduced the time and expense from model-building from weeks to minutes, allowing Noodle to generate dozens of plausible models based on a university’s data. Models can then be compared, assessed for implementation practicality, and fine-tuned. There are no longer dead-ends or the sunk cost fallacies that plague sophisticated models built across higher education in the past. If a model is reductive beyond usefulness, it can be discarded or revamped quickly, allowing us to find methods that give critical insights and to continually fine-tune those methods over time. 

Pricing

For partners already using N. Manage, the good news is that much of the heavy lifting is done already. Noodle has already aggregated and normalized data from across the student lifecycle at a grain size that is ideal for data modeling. Running a university’s data through the AI-enhanced modeling software and completing the process described above becomes a function of the complexities of the questions we set out to answer and the sophistication of the required modeling.