Real-Time Prediction of Student’s Locality towards Information Communication and Mobile Technology: Preliminary Results
Chaman Verma1, Zoltán Illés2, Veronika Stoffová3
1Chaman Verma, Department of Media and Educational Informatics, Eötvös Loránd University, Budapest, Hungary.
2Zoltán Illes, Department of Media and Educational Informatics, Eötvös Loránd University, Budapest, Hungary.
3Veronika Stoffová, Department of Mathematics and Computer Science, Trnava University, Trnava, Slovakia.
Manuscript received on 18 April 2019 | Revised Manuscript received on 24 May 2019 | Manuscript published on 30 May 2019 | PP: 580-585 | Volume-8 Issue-1, May 2019 | Retrieval Number: F2566037619 /19©BEIESP
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: The present paper used supervised machine learning to predict the student’s locality towards Information and Communication Technology (ICT) and Mobile Technology (MT) of Indian and Hungarian University. For this, a primary dataset is used with 331 instances and 38 features which are related to the 4 major ICT parameters belongs to the attitude, development and availability, educational benefits and usability of modern ICT resources and mobile technology used in education. To predict the locality, three machine learning classifiers multi-layer perceptron (ANN), K-nearest neighbor (KNN or Idk) and Random Forest (RF) are used with hold out method, Leave One Out and K-Fold cross-validation methods. Further, to enhance the prediction accuracy, RF used CorrelationAttributeEval, ANN used InfoGainAttributeEval and KNN used OneRAttributeEval Feature Selection techniques. The outcome of the study reveals that the Feature Selection algorithm significantly improved the prediction accuracy of each classifier. To compare the accuracies of the extracted dataset, T-test at 0.05 significant level was also used. T-test did not find a significant difference between RF and KNN towards CPU training time and another hand, a significant difference is found between ANN and KNN; ANN and RF classifier. It is also proved that KNN classifier has outperformed others in stabilized accuracy and induced optimum time in locality prediction of students.
Index Terms: Classifier, Feature Selection, Locality Prediction, Leave One Out, K-Fold, Hold Out, Real-Time.
Scope of the Article: Agricultural Informatics and Communication