IoT System for School Dropout Prediction Using Machine Learning Techniques Based on Socioeconomic Data

• Main Authors: Francisco A. da S. Freitas, Francisco F. X. Vasconcelos, Solon A. Peixoto, Mohammad Mehedi Hassan, M. Ali Akber Dewan, Victor Hugo C. de Albuquerque, Pedro P. Rebouças Filho
• Format: Article
• Language: English
• Published: MDPI AG 2020-10-01
• Series: Electronics
• Subjects: IoT; school dropout; machine learning
• Online Access: https://www.mdpi.com/2079-9292/9/10/1613

School dropout permeates various teaching modalities and has generated social, economic, political, and academic damage to those involved in the educational process. Evasion data in higher education courses show the pessimistic scenario of fragility that configures education, mainly in underdeveloped countries. In this context, this paper presents an Internet of Things (IoT) framework for predicting dropout using machine learning methods such as Decision Tree, Logistic Regression, Support Vector Machine, K-nearest neighbors, Multilayer perceptron, and Deep Learning based on socioeconomic data. With the use of socioeconomic data, it is possible to identify in the act of pre-registration who are the students likely to evade, since this information is filled in the pre-registration form. This paper proposes the automation of the prediction process by a method capable of obtaining information that would be difficult and time consuming for humans to obtain, contributing to a more accurate prediction. With the advent of IoT, it is possible to create a highly efficient and flexible tool for improving management and service-related issues, which can provide a prediction of dropout of new students entering higher-level courses, allowing personalized follow-up to students to reverse a possible dropout. The approach was validated by analyzing the accuracy, <i>F1 score</i>, <i>recall</i>, and <i>precision</i> parameters. The results showed that the developed system obtained 99.34% accuracy, 99.34% <i>F1 score</i>, 100% <i>recall</i>, and 98.69% <i>precision</i> using Decision Tree. Thus, the developed system presents itself as a viable option for use in universities to predict students likely to leave university.