An Improved Model to Predict Student Performance using Teacher Observation Reports

Abstract

Predicting students’ performance is a highly discussed problem in educational data mining. A tool that can accurately give such predictions would serve as a valuable resource to teachers, students, and all educational stakeholders as it would provide essential insights. Students can be further guided and fostered to achieve their optimal learning goals. In this paper, we propose an improved method to predict students’ performance in entrance examinations using comments that their cram school teachers took throughout lessons. Teachers in these cram schools observe their students’ behavior closely and give reports on the efforts taken in their subject material. We compare our previous model with a new and improved one to show that teachers’ comments are qualified to construct a reliable tool capable of predicting students’ grades efficiently. These methods are new since studies previously focused on predicting grades mainly using student data such as their reflection comments or earlier scores. Our improved experimental results show that using this readily available feedback from teachers can predict students’ letter grades with an accuracy of 68%.