Student retention is a persistent challenge that educational institutions face, with repercussions ranging from reduced funding to negative educational outcomes for students.
Traditional approaches to tackle this issue often lack the granular insights needed for effective early intervention. The one-size-fits-all solutions commonly employed do not consider the varying degrees of academic performance and engagement levels among students.
How can educational institutions adopt a more targeted, data-driven approach to identify at-risk students who are highly engaged but not academically successful, and then provide them with a personalized curriculum that can significantly improve retention rates?
Data Sources
We will be using 3 sources of data to create a personalized student retention platform:
Academic Data
This dataset captures the academic performance of each student. Metrics such as Overall GPA, Recent Grade Trends, and Attendance Rate are included.
It might have a student's overall GPA, the pattern in grades (rising, declining, stable), and how often the student attends classes.
{
"student_id": 1,
"gpa": 3.5,
"subject": "math"
}Engagement Data
This dataset focuses on a student's engagement with the course or courses they are enrolled in. Metrics like Forum Participation and Time Spent on the Course Platform give a nuanced understanding of how actively a student is participating in the learning process.
It may capture how often a student participates in online course discussions or the average time spent on course materials.
{
"student_id": 1,
"forum_participation": "High",
"time_spent_on_platform": "Above Average"
}Course Curriculum
This dataset provides information about the course itself. It will contain details such as the Subject of the course, a Description outlining what the course covers, a Link to the course materials or syllabus, and the Expertise Level required or recommended for the course (Beginner, Intermediate, Advanced).
{
"Subject": "Mathematics",
"Description": "algebra, calculus, and geometry",
"Link": "http://www.example.com/mathematics_course",
"Expertise_Level": "intermediate"
}Chain
Identify At-Risk Students
This is the first stage where we analyze both the Academic and Engagement Data to flag students who are potentially at risk of dropping out or failing.
Metrics Used:
- Overall GPA
- Recent Grade Trends
- Attendance Rate
How It Works:
We use predetermined criteria to identify students who may be struggling academically. For example, students with an overall GPA below a certain threshold or a declining grade trend might be flagged.
Output:
The output is a list of students who are identified as 'At-Risk' based on the Academic and Engagement Data.
Filter Amongst Those, the Highly Engaged
In this stage, we further narrow down our list by identifying which of the at-risk students are highly engaged in the course but are not performing well academically.
Metrics Used:
- Forum Participation
- Time Spent on Course Platform
How It Works:
Here, we filter the list of at-risk students to focus on those who are showing high engagement but are still struggling. The rationale is that these students are putting in the effort but may need targeted support to improve their academic performance.
Output:
A curated list of students who are at-risk yet highly engaged, and potentially most receptive to interventions.
Generate a Personalized Curriculum with Links
In the final stage, we use the Language Learning Model (LLM) to generate a personalized curriculum for the highly engaged but at-risk students.
Metrics Used:
- Subject
- Expertise Level
- Course Description
- Link to Course
How It Works:
The LLM takes the Academic Data, Engagement Data, and Course Curriculum Data to generate personalized course recommendations with links. It aims to match the student's needs and engagement level with the most suitable courses.
Output:
The output is a detailed curriculum guide for each student that includes recommended courses, descriptions, and direct links to those courses.
