Scribd
Upload
12 views2 pages

SFA paper 6

This research presents a machine learning approach to predict student performance in degree programs, addressing challenges like student diversity and course relevance. A bilayered model utilizing ensemble predictors and probabilistic matrix factorization was developed and tested on UCLA data, demonstrating improved prediction accuracy. The method has potential applications in personalized education, curriculum design, and institutional policy strategies to enhance graduation rates.

Uploaded by

sahilchakrabortty912
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
12 views2 pages

SFA paper 6

This research presents a machine learning approach to predict student performance in degree programs, addressing challenges like student diversity and course relevance. A bilayered model utilizing ensemble predictors and probabilistic matrix factorization was developed and tested on UCLA data, demonstrating improved prediction accuracy. The method has potential applications in personalized education, curriculum design, and institutional policy strategies to enhance graduation rates.

Uploaded by

sahilchakrabortty912
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 2

# PAPER – 6 (A Machine Learning Approach for Tracking and Predicting

Student Performance in Degree Programs)


Abstract
This research focuses on predicting student performance in degree programs
using machine learning. Traditional approaches focus on course or problem-
level performance, while this study addresses challenges specific to degree
programs: student diversity, course relevance, and evolving academic progress.
A bilayered machine learning model was developed using ensemble predictors
and course clustering through probabilistic matrix factorization. The approach
was tested on UCLA data, showing superior prediction accuracy compared to
existing methods.

Introduction
Student loan debt in the U.S. has surpassed $1 trillion, partly due to prolonged
graduation times. Ensuring students graduate on time requires predicting their
performance accurately and intervening early. While predicting performance at
the course or problem level has been studied, predicting overall program
outcomes introduces challenges such as diverse student backgrounds, variable
course sequences, and the need to track academic progress continuously. This
study proposes a novel method addressing these challenges to help students
succeed and manage their academic journey effectively.

Methodology
The study introduces a bilayered prediction model:
1. Base Layer: Predicts course grades using advanced machine learning
algorithms based on students' evolving performance.
2. Ensemble Layer: Aggregates predictions from the base layer to provide
final predictions. This layer uses historical and current data for better
accuracy.
The model incorporates:
 Course Clustering: Uses latent factor models and probabilistic matrix
factorization to group related courses and reduce noise.
 Progressive Prediction: Continuously updates predictions as new data
becomes available, integrating student progress over time.
Data from UCLA's Mechanical and Aerospace Engineering department was
used to evaluate the model.

Technology Used
1. Machine Learning Models:
 Logistic Regression
 Random Forest
 k-Nearest Neighbors (kNN)
 Linear Regression
2. Probabilistic Matrix Factorization: To discover relationships among
courses and create clusters.
3. Ensemble Learning: For combining predictions from multiple models
progressively.

Future Scope
The proposed method has potential applications in:
 Personalized Education: Tailoring academic advice and interventions
based on predictions.
 Curriculum Design: Informing which courses or sequences need
restructuring.
 Expansion to Elective Courses: Incorporating elective course predictions
to provide holistic recommendations.
 General Policy Design: Guiding institutional strategies to improve
graduation rates and reduce costs.

You might also like