An Efficient Cluster-Based Feature Selection and Classification Framework for Student Dropout Prediction

Abstract

Massive Open Online Courses (MOOC) is an extensive way of providing online education to the students all over the world. According to statistics, this educational system has millions of students enrolled in hundreds of courses across a variety of activities. Since its inception, MOOC has faced a number of issues, one of which is known as the "student dropout prediction accuracy," which is also a significant difference between traditional teaching and MOOC. As a result of this fact, MOOC's overall performance has a negative impact on the true purpose of distance learning. In MOOCs, however, the gap between course registration and course completion is quite large. On the plus side, emerging technologies have opened up several opportunities for students to receive education online; however, due to a variety of factors, the dropout rate of online students is higher than that of traditional school students. The goal of this study is to better understand and predict the MOOC dropout rate. The multiple models and evaluation metrics generating variety of results as extracted from literature review. In this model, a hybrid cluster based feature selection model is implemented in order to optimize the class prediction. In this model, a hybrid cluster based metaheuristic model is designed and implemented on the classification problem.