Interpolated Contractive Auto Encoder Sequential AI For Student Academic Performance Prediction

Abstract

Education is the foundation of individuals for encouragement intellectual development as well as determining prospect leaders.  Student performance forecast an essential in edifying institutions, giving precious insights to student’s academic results as well as assisting educators at executing adapted intercession. In preceding works, numerous variants of ML as well as AI methods have executed by increased accuracy. Student performance forecast espouse prognostic method to predict as well as compute student prospect educational result depend on historic information as well as pertinent features gathered over classes of syllabus. But, lack of clearness because of assorted nature of features which escort to biases, mandatory disparity or even demands at inferring method insinuation not mentioned. To mention above problems, novel method termed Interpolated Contractive Auto Encoder Sequential AI (ICAESqAI) Model is introduced. Major aim objective of ICAESqAI Model is to carry out effective predictive analytics on collected student data samples in educational institutions. ICAESqAI Model consists of four steps, namely data acquisition, preprocessing, FS as well as categorization to enhance student performance forecast. Initially, the student data samples are gathered as of database in data acquisition phase. Followed by, the data pre-processing is carried out using Nevanlinna–Pick interpolation estimator for handling the missing data and Berk-Jones test for outlier data removal from the input dataset. After data pre-processing, FS is performed using Contractive Auto Encoder to choose pertinent features as of the input dataset based on Marczewski-Steinhaus similarity index. After feature selection process, the classification process is carried out using sequential AI model with tamas correlation coefficient function. The fine tuning process is carried out for minimizing the error and enhancing the student performance prediction accuracy. Experimental assessment is conducted through dissimilar evaluation parameters. Outcome denote ICAESqAI attained higher accuracy in student performance prediction with lesser time consumption compared to existing deep learning methods.