Non-cognitive skills : the art of learning to learn well (Part C)
Date of Issue2016
School of Electrical and Electronic Engineering
This project deals with the application of data analytics to education, particularly students’ knowledge acquisition process. The target group is students using e-learning platforms. This project involves assessing the mastery level of students for various knowledge skills through machine learning techniques. Bayesian Knowledge Tracing, which is built upon the Hidden Markov Model, is the algorithm being incorporated to develop a model to gauge students’ mastery progression. This algorithm revolves around the manipulation of four main probabilities to calculate every student’s level of knowledge and learning transition for every skill for every attempt made. The method of Brute Force Search is implemented to generate exhaustive combinations, with which two error metrics are utilized to obtain the best fitting parameters to serve as the input for every student for every skill. The results plotted are subsequently regularized to smoothen the errors so as to obtain an accurate reflection of a student’s ability to grasp a skill after several attempts over time. R Studio is being used for implementation of this algorithm. Real-life data from students of Palmview Primary School who have completed several exercises through an e-learning platform is used as input to the model to generate the individualized knowledge tracing results for every student for every skill. Inferences of students’ behaviour and psychological patterns are made to classify different types of students. As such, the application of data analytics to education has great benefits of helping not only students to reflect on their ability, but also teachers to understand their students better.
Final Year Project (FYP)
Nanyang Technological University