Age Group Predictive Models for the Real Time Prediction of the University Students using Machine Learning: Preliminary Results

This paper is written to support the real time prediction of the age group of the students towards four different ICT parameters such as the development and availability of modern ICT Resources (DAICT), Attitude of students towards ICT and mobile technology in education (AICTM), Use of ICT and mobile technology in education (UICTM) and educational benefits of ICT and mobile technology in education (EBICTM) in Hungarian and Indian University. The investigator has collected the primary dataset during the academic year 2017-18 from two countries. The present study formulates the idea of feature aggregation significantly improved the prediction accuracy of the age group of students towards ICT parameters. The authors solved a multiclassification problem using tested primary dataset with 5 popular supervised machine learning classifiers such as K-nearest neighbor (KNN), Random forest (RF) and Support vector machine (SVM), Bayesian network (BN) and decision tree (C5.1) in SPSS IBM Modeler version 18.1. The concept of aggregation of features provides maximum accuracy of 91.4% is provided by joint classifiers (RT, SVM, and BN) in the prediction of age against 37 features. The second winner group (RT, SVM, C5.1) provides 79.2% accuracy only with 16 features. Although, aggregation of features has significantly increased the accuracy in prediction of student's age group towards four ICT parameters instead of an individual. Findings of the study also stabilized dataset with 337 instances with 37 features in the prediction of the student's age group belongs to India and Hungary. This paper may be play a significant role in development and support of real time age prediction of the student's age at individual and collective parameters of ICT in both countries.