Student Attrition Analysis
- Python
- SQL
- Databricks
- →~28,000 student records, 6 schools, 20+ courses, 7 academic years
- →Year 1 identified as the single largest share of all departures
- →Presented to a senior leadership panel; feeding a live retention decision
Context
Attrition analysis is a standing responsibility for our team, produced for academic review panels in past cycles. This year we were asked to run it again and present the findings directly at the panel meeting.
The work
I pulled the student data from Databricks: roughly 28,000 records across six schools and more than twenty courses, spanning seven academic years. I used Python and SQL to shape the raw data into analysis-ready tables, then used AI-assisted analysis to surface trends and dug into what was driving each one. From there I built the narrative storylines for the presentation. The most significant was that Year 1 accounts for the single largest share of all student departures across the seven-year window, which makes it the highest-impact point to intervene.
Pitching it to leadership
The findings went into a 19-slide deck built in PowerPoint and presented to the panel. Ahead of the meeting I ran it through several review rounds with senior stakeholders, a mix of substantive challenges to specific findings and framing and lighter presentation polish, so the deck was both analytically sound and pitched right for the audience.
Impact
Since the presentation the analysis has fed directly into a specific retention-programme decision, giving the Year 1 finding a concrete downstream effect on how the institution shapes its retention strategy.