Aligning Learners' Expectations and Performance by Learning Analytics Systemwith a Predictive Model
Project Overview
The document explores the integration of generative AI in education, specifically through the implementation of a Learning Analytics (LA) system designed to foster self-regulated learning in higher education. It details the methodology for gathering student expectations about LA, utilizing both qualitative and quantitative approaches, such as the Student Expectations of Learning Analytics Questionnaire (SELAQ) and focus groups. The LA system aims to provide early grade predictions and insights into student learning behaviors, empowering learners to adjust their strategies for improved academic outcomes. Initial evaluations of the system reveal positive feedback from students concerning its usability and effectiveness, indicating that generative AI can play a significant role in personalizing learning experiences and enhancing academic performance. Overall, the findings underscore the potential of generative AI technologies to transform educational practices by supporting student engagement and self-directed learning.
Key Applications
Learning Analytics System with Predictive Model
Context: Higher education, specifically in Slovenian engineering education for students at the bachelor, master, and PhD levels.
Implementation: The system was implemented using a user-centered approach guided by student requirements, involving the SELAQ questionnaire and focus groups to gather expectations. The interface was developed using HTML and Javascript, and the predictive model was developed in Python.
Outcomes: The system provides feedback on learning patterns, early grade predictions, and comparisons with peers, helping students to self-regulate their learning processes. Initial evaluations showed a high precision in identifying at-risk students and positive user feedback.
Challenges: Challenges include the accuracy of predictions, potential anxiety caused by predicted grades, and the need for ethical considerations in data usage.
Implementation Barriers
Ethical
Potential anxiety and stress related to providing students with predicted grades.
Proposed Solutions: An ethical board review is planned to address these concerns and ensure responsible data usage.
Technical
The model's predictions may have false positives, leading to incorrect passing grade predictions.
Proposed Solutions: Cross-validation techniques were applied to improve prediction accuracy, but further refinement and testing are needed.
Implementation
The current system is limited to one course, limiting its scalability and applicability.
Proposed Solutions: Future work includes expanding the system to other courses and incorporating feedback from additional stakeholders.
Project Team
Saša Brdnik
Researcher
Boštjan Šumak
Researcher
Vili Podgorelec
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Saša Brdnik, Boštjan Šumak, Vili Podgorelec
Source Publication: View Original PaperLink opens in a new window
Project Contact: Dr. Jianhua Yang
LLM Model Version: gpt-4o-mini-2024-07-18
Analysis Provider: Openai