Using Neutrosophic Soft Set to predict Higher Education Academic Performance

Sally Afchal^1,*, Muhammad Eid Balbaa²

¹College of Business Administration, American University of the Middle East, Kuwait

²Tashkent State University of Economics, Uzbekistan

Emails: sally.afchal@aum.edu.kw; M.balbaa@tsue.uz

Abstract

Neutrosophic Logic is a neonate research field in which every proposition is assessed to have the proportion (percentage) of truth in a sub-set T, the proportion of indeterminacy in a sub-set I, and the proportion of falsity in a sub-set F. Neutrosophic set (NS) is effectively implemented for undetermined data processing and establishes benefits for handling the indeterminacy data. In the academic industries, early performance prediction of students is significant to the academic community so strategic interference might be planned before students attain the final semester. Forecasting the performance of students has turned into a challenging task owing to the rising number of data in educational procedures. The educational data mining (EDM) models are involved in extracting a pattern to explore hidden data from educational information. Currently, Machine learning (ML) and Artificial intelligence (AI) are implemented in numerous domains generally in the field of education to evaluate and analyze several features of educational datasets gathered from many educational institutions. This study develops a Leveraging Generalized Possibility Neutrosophic Soft Set with Feature Selection for Accurate Students’ Academic Performance Prediction Model (GPNSSFS-SAPPM). The intention of the proposed GPNSSFS-SAPPM system relies on improving the prediction model of students’ higher education performance using metaheuristic optimization algorithms.  The data pre-processing model is employed at first by applying mean normalization for converting input data into a suitable format. In addition, the golf optimization algorithm (GOA) is exploited for the feature selection process. Followed by, the classification process is done by generalized possibility neutrosophic soft set (GPNSS). At last, the parameter tuning process is performed through henry gas solubility optimization (HGSO) algorithm to improve the classification performance of the GPNSS classifier. A wide-ranging experimentation was performed to prove the performance of the GPNSSFS-SAPPM method. The experimental results specified that the GPNSSFS-SAPPM model underlined advancement over other recent techniques.

Keywords:  Generalized Possibility Neutrosophic Soft Set; Neutrosophic Logic; Students’ Academic Performance; Feature Selection; Henry Gas Solubility Optimization