Bu araştırmada Akdeniz Üniversitesi Eğitim Fakültesi Sınıf Öğretmenliği
Bölümü’nden 2012-2017 yılları arasında mezun 578 öğrencinin cinsiyet, medeni durum,
kayıt yaşı ve 1. sınıf 1. dönem ara sınav puanları verisi kullanılarak
mezuniyet notlarını tahmin etmek için modeller oluşturulmuştur.
Anonimleştirilmiş öğrenci verisi üzerinde çoklu doğrusal regresyon analizi ve
yapay sinir ağları ile oluşturulan modellerin çapraz değerlendirme sonuçlarında
öğrenci başarısını kestirmede birbirine yakın sonuçlar verdiği görülmüştür.
Ortalama mutlak hata yüzdesi değerleri baz alınarak yapılan değerlendirmede
regresyon analizi modeli %94.30 başarı sağlarken yapay sinir ağları modeli %94.43
başarı sağlamıştır. Modeller öğrenci başarısına etki eden faktörlerin
ağırlıklarını belirlemek için bulgular sağlamıştır. Araştırma kapsamında
geliştirilen modellerin etkili olduğu eğitimde toplam kalite yönetimi
bağlamında sıfır hata hedefinde proaktif pozisyon alarak öğrenci başarısını
artırma gibi çalışmalarında kullanılabileceği söylenebilir.
Akdeniz Üniversitesi BAP Birimi
SDK-2018-4077
In this research,
models were developed to estimate the graduation grades of 578 students who
graduated from the Department of Primary Education, Faculty of Education,
Akdeniz University between 2012-2017 by using the data related to gender,
marital status, age of enrollment and midterm exam points of 1st semester of
1st grade. It has been seen that the models formed by regression and artificial
neural networks on anonymized student data give close results in predicting
student success as a result of cross-evaluation. In the evaluation based on
Mean absolute percentage error values, the regression model achieved 94.30%
success and the artificial neural networks model achieved 94.43% success. The
models provided findings to determine the weights of the factors affecting
student achievement. It can be said that the models developed with in the scope
of the research can be used in studies such as zero error in the context of
total quality management, taking proactive positions, increasing student
achievement and guiding.
SDK-2018-4077
Primary Language | Turkish |
---|---|
Subjects | Studies on Education |
Journal Section | Educational Sciences and Sciences of Field Education |
Authors | |
Project Number | SDK-2018-4077 |
Publication Date | December 31, 2019 |
Published in Issue | Year 2019 Volume: 10 Issue: 3 |
This journal uses a CC BY-NC-SA license.
editor@e-ijer.com http://www.e-ijer.com Address: Ege University Faculty of Education İzmir/Türkiye