Data analytic on student assessment results

• Main Author: Zhou, Liqin
• Other Authors: Shum Ping
• Format: Final Year Project
• Language: English
• Published: 2019
• Subjects: Engineering::Electrical and electronic engineering
• Online Access: http://hdl.handle.net/10356/78761
• Institution: Nanyang Technological University

Multiple educational institutes have adopted e-learning platform to enhance their education systems. A significant amount of student learning data has been generated from student activities on the e-learning platform. Many universities tried to make fully utilize these educational data to track student performance and improve the quality of education. This motivates the proposal for this final year project. In the first phase of this project, machine learning techniques are implemented to identify low-participation students and predict students’ final exam performance based on their continuous assessment as well as their engagement with the virtual learning environment (VLE). The VLE is a new form of e-learning system that stores materials, course lectures, and assessment. By only capturing the first few weeks’ student records, Course coordinates can view the prediction result and identify students at risk and take further actions in time. The machine learning implementation stage consists of data preparation, data preprocessing, feature engineering, correlation analysis, data modeling, model tuning, and performance evaluation. In the second phase of this project, a data analysis website is constructed for student statistics visualization purpose. Outcome-Based Learning is implemented. A weekly outcome gain mapping for the designated student can be generated on the website as well. The website development contains two part: the database design and website design. The Open University (OU) is a large university located in Europe. Nearly 200,000 students participated in various courses in the OU. After noticing the transferable properties and structures between OASIS data and OU data, this project implements both simulated OASIS data and real-world student data to avoid limitations of simulated data. OASIS data was used for the website visualization and the OU dataset was implemented for Machine Learning prediction.