EduBot: a proof-of-concept for a high school motivational agent

Abstract

Motivation appears to plays a role in student dropout. This paper uses a dataset with high school information (756 entries) for creating a model that relates year failures with the other features on the dataset. The resulting model, based on XGBoost, shows that the top-ranked features are related with motivation issues corroborating with the initial observation. In this sense, a digital assistant embedded with a motivation module may aid on improving motivation and on avoiding dropout. In other words, if the predictor detects year failure possibility it start to act on motivating the student. Considering this paper being direct to underage students, this paper stops on the proof-of-concept level considering the actual dialogues and live tests are expected to performed with the support of educational psychologists.