ANALYTICS | CASE STUDY

Predict Higher Education Student Retention Risk

COMPANY OVERVIEW

A community college providing open access to quality higher education was looking for a way of predicting student retention risk so they could intervene and help at-risk students succeed.

BUSINESS NEED

The college’s student retention rate over 6 years was below 50 percent. They knew they needed to find proactive ways of identifying at-risk students, but weren’t sure how to best leverage their data to do it. The college’s Institutional Research team brought in Ironside to test predictive modeling techniques, evaluate and select a set of predictor variables on which to base student risk scores, segment students into groups based on GPA attainment and retention likelihood, integrate with the college’s student services model, data warehouse, and BI platform, and output individualized student risk scores and intervention/retention strategies for at-risk students.

SOLUTION

The Ironside Data Science & Advanced Analytics team is dedicated to helping organizations realize the full potential of their data resources. The client brought in the team to do the following:

Define the Vision

Collect data at a granularity that permits risk score creation for each enrolled student
Trigger actions and alerts based on student behaviors or academic events
Fit recommendations to each student’s needs, allowing long-term intervention plans
Feed into the data warehouse to compare with actuals/the student services model

Prepare the Data

Client needed to prepare and transform their data for use in predictive models
Analysts spent 70-80% of their total project time working with Ironside experts to extract and refine the information needed for their predictive model

Develop & Refine Scoring Methods

Isolate productive variables and designate variable categories for each model
Train models against 10 years’ worth of student data (approx. 68,000 unique records) using multiple predictive algorithms
Implement linear regression models based on findings from modeling tests
Weight chosen variables based on relative importance in overall risk score outputs
Identify four student service quadrants: GPA & retention safe, GPA safe but retention at risk, GPA at risk but retention safe, and GPA & retention at risk
Feed results into data warehouse and begin defining next best actions based on reports

Ensure Future Progress

Client’s internal resources gained hands-on experience with their student retention risk models and with variable and algorithm selection
Ensured they could use their student retention risk models as a springboard into other modeling projects
Opened the door for future insights into student behavior and other operational projections like donor optimization

KEYS TO SUCCESS

Through predictive analysis, the client was able to:

Enable mass deployment of individualized student services and target first year experiences
Refine the views on established business practices with hard facts and accurate predictions, fueling change

Affirm the importance of specific variables for student success and tie them to actual effects on the student population
Empower analysts with a predictive analytics knowledge base so they can better improve policies and assist students

READ ADDITIONAL CASE STUDIES

Strategy Data Analytics Platform

Empower the Team with Experience Design

Enable the Tableau Enterprise

Formulate a Data & Analytics Roadmap for Acquisition Growth Strategy

Establish a Collaborative Data Governance Strategy

Build an Enterprise Data Warehouse for Self-Service Analytics

Implement a Single View of the Customer

Enable Current and Future Growth with Adaptive Insights

Enable Interaction Reporting with Tableau

Use Data to Drive Personalized Marketing

Allow Flexibility & Resiliency with Standardized ETL

Migrate from Oracle Discoverer to Cognos Analytics

Enable Cognos Analytics Migration in Financial Services

Become one of our successful clients.

Take our Data & Analytics Self-Assessment

Send this to a friend

Your emailRecipient email