Educational Data Mining: Predicting Candidates’ Placement Status in Physical Education and Sports Education Program

Abstract

Educational data mining’s primary purpose being to extract useful information from educational data in order to support decision-making on educational issues. One of the most preferred methods in educational data mining is prediction. The primary purpose of the current study is to predict whether or not candidates will be admitted into the PESE program according to different algorithms. Within the scope of this research, data was obtained from 1,671 candidates who applied to join the PESE program of a state university in Turkey between 2016 and 2020 were studied. The Random Forest, kNN, SVM, Logistic Regression, and Naïve Bayes algorithms were each used to predict whether or not a candidate could admit to the PESE program. According to the findings, the algorithms’ classification accuracy from highest to lowest is Random Forest (.985), SVM (.845), kNN (.818), Naïve Bayes (.815), and Logistic Regression (.701), respectively. In other words, the Random Forest algorithm is shown to have correctly classified the instances almost exactly. Other findings from the study are discussed in detail, and suggestions put forth for future research.