Leveraging Data Analytics in Science Education: Bridging the Gap between Computer Science and Pedagogical Practices

Abstract

The integration of data analytics into science education has led to a revolutionizing of pedagogical practices through data-driven decision making, personalized learning, as well as prediction of student performance. In this research, machine learning algorithms such as Decision Tree, Support Vector Machine (SVM), Random Forest, and K-Means Clustering are applied to student data to primarily seek to analyze student data and improve the learning strategies. It is evident from the experimental results that Random Forest shows the maximum accuracy 92.3%, SVM acquires 89.7% and Decision Tree attains 85.6%. Using K-Means Clustering to group students into performance-based clusters yielded interpretations into learning behaviors. This approach demonstrated (i) 7.8% better prediction accuracy than related studies and (ii) 15% higher efficiency in categorizing at risk students. The research findings confirm that data analytics significantly impacts on educational effectiveness and that it supports educators making educated interventions. Nevertheless, more needs to be explored in terms of challenges in algorithmic bias and data privacy. By doing this, this study bridges the gap between education and computer science, to show that machine learning has the potential of optimizing student learning outcomes. Further improvement in the predictive capabilities in the future can be achieved using deep learning models and real time adaptive learning systems.