Novel ANN based regression and improved Lion Optimization Algorithm for efficient prediction of student performance

Abstract

Prognosticating the performance of students have become challenging because of huge amount of data in pedagogical databases. But this prediction will afford the teachers with a pro-active opportunity to come up with additional resources for learners to enhance their probabilities of enhancing their grades. Thus the academic performance of students have to be predicted to aid a teacher to find the struggling students easily. To select relevant data by using the new improved lion optimization (ILO) algorithm and to predict the student’s performance and dropout analysis via the novel ANN based regression (ANN-R). At first, the population and parameters are initialized. Then the initial best agent is defined based on FF (Fitness Function) and the position of CSA (Current Search Agent) is updated. When the iteration ends, the optimized result is obtained and then data selection is performed using new ILO algorithm. Subsequently, the train test splitting is performed to predict the performance of the student and dropout analysis by using novel ANN-R. Finally, the dropout analysis is found based on fees structure, student as well as attendance performance. The performance of the introduced methodology is compared with traditional algorithms such as CNN (Convolutional Neural Network), SVM (Support Vector Machine) and GOOGLE-NET. The proposed methodology is also analyzed with respect to Mean Absolute Error (MAE), Mean Squared Error (MSE) and Root Mean Squared Error (RMSE). The analytical results explored that the proposed system shows minimum MAE (1.050008461), RMSE (1.334106084) and MSE (1.779839044) value than the existing system which indicates the low prediction errors thereby enhancing the accuracy of the proposed system.