An Optimized Hybrid Deep Learning & Ai Enabled Scheme For Student’s Academic Performance Prediction: Educational Data Mining

Abstract

Education is regarded a crucial one for productive life that offers resources needed. With the technology advancement like artificial intelligence, higher education institutions incorporate this technology to traditional teaching models. The prediction of academic success gained interest in education as a strong academic record thereby improving the university’s ranking and thus enhances the employment opportunities of the students. Most of the modern learning institution faces some challenges on analysing performance thus providing higher-quality education, and formulating strategies to estimate the performance of student and to identify the future needs. Thus, educational data mining (EDM) is employed for automating the prediction system of academic performance. In this work, prediction and classification of student’s academic performance is carried with the use of intelligent deep learning-aided scheme. The student dataset is pre-processed primarily for removing unwanted or redundant data after which the data is clustered using Hybrid Fuzzy C-Means (FCM) & Expectation Maximization (EM) model. Deep neural network (DNN)-based feature extraction is performed and optimal selection of feature is made with the use of Chaotic Shark Smell Optimization (CSSO) approach. After this, the prediction of academic performance of student is performed with the use of Hybrid Inception V3 & SqueezeNet classifier model and the data predicted is stored in cloud for further analysis and monitoring. Finally, the performance analysis is carried on various metrics so as to validate the performance of proposed scheme.