Predictive model to support knowledge generation: The case of outcome-based education assessment in Malaysia.

Visual analytics increases human cognition's capabilities through data exploration. Data exploration to further understand data that was produced was made possible by coupling interactive visualization with analytics which became known as visual analytics. With the increasing data production in everyday lives, there has been a notable rise in visual analytics research. However, there has yet to be any research that investigates the skills attained by students outside of the classroom and linking that to the learning outcomes set by the schools, faculties, or universities. This paper reports a phase of the study that was done in discovering the predictive the relationship between students' outcome-based education results with their out of classroom activities using visual analytics. In order to design the visual analytics system, there is a need to perform research on the suitable interactive visualization and data analysis to achieve the expected outcome. This paper reports the study that was done in discovering the predictive modelling technique that yields the highest accuracy result. This study makes use of students' semester results and the recorded out-of-classroom activities data of theirs from Universiti Malaysia Sabah' Accounting program. The dataset was tested with four algorithms which are Random Forest, Gradient Boosting, Neural Networks, and Support Vector Machine. Results showed that Random Forest yield the highest accuracy result. [ABSTRACT FROM AUTHOR]