Transforming Learning Data into a Machine Learning Model to Help STEM Students Transition to University

EasyChair Preprint 10216

Abstract

Previous research has applied Machine Learning (ML) to predict student success in higher education using entry data and cumulative GPA scores. Our research aims to add student learning and performance data in specific STEM courses to the ML modelling process. In the initial phase, the data included self-report scores on inventories that assess students’ learning strategies, metacognitive awareness, mindset, and misconceptions about how the brain works, as well as learning analytics and course grades. This data is collected as part of an orientation program that aims to develop students’ self-regulated learning capabilities. Our goal is to provide evidence to inform this program, use the results to predict student success and challenges in first-year STEM courses, and inform proactive help for students’ transition to university. This paper provides a step-by-step introduction to the methodology used to build a prototype of the ML model underpinning this research and future directions.

Keyphrases: AI for Education, Data Mining Technique, Decision Tree, Predicting Student Performance, Random Forest, STEM education, Students' Performance Prediction, elastic net classifier, machine learning, science education, self-regulated learning, student learning, student performance prediction, time series

@booklet{EasyChair:10216,
  author    = {Neeraj Katiyar and Armin Yazdani and Janette Barrington and Kira Smith and Valerie Bourassa and Hilary Sweatman and Marcy Slapcoff},
  title     = {Transforming Learning Data into a Machine Learning Model to Help STEM Students Transition to University},
  howpublished = {EasyChair Preprint 10216},
  year      = {EasyChair, 2023}}