Predictions of academic achievements of vocational and technical high school students with artificial neural networks in science courses (physics, chemistry and biology) in Turkey and measures to be taken for their failures

¹ Karamanoğlu Mehmetbey University, Institute of Science, Karaman, 70100, Turkey
² Karamanoğlu Mehmetbey University, Education Faculty, Department of Primary Education, Karaman, 70100, Turkey

^a Corresponding author: ayagci89@gmail.com

Abstract

The aim of this study is to predict academic achievements of vocational and technical high school (VTHS) students with artificial neural networks (ANN) in physics, chemistry and biology courses in Turkey and reveal measures to be taken for their failures. The study group consisted of 922 students studying in 10^th and 11^th grade in VTHS. This study was conducted with the survey method and a 34-item demographic questionnaire was developed in order to collect the data. The parameters in the questionnaire were identified as the items that were considered to influence academic achievements of the students. Opinions of 3 field specialist, 1 measurement and evaluation specialist and 2 technical teachers were taken and it was supported by the literature for the content validity and the KR20 reliability coefficient was found as .90 using SPSS 16.0 package program. The items in the questionnaire that are considered to influence academic achievements of the students were approved as independent variables/input and academic achievement mean scores of the students in physics, chemistry and biology courses in the previous year were approved as dependent variables/output. Academic achievements of the student were predicted with ANN in Matlap R2016a program using these parameters and a model was created. A successful academic achievement prediction system with an average sensitivity of %98 was developed over 922 data and the measures to be taken to prevent the failures of the students were determined at the end of the study.