Prediction of Higher Education Institutions Examination (YKS) Scores with Artificial Intelligence Techniques

Betül Ersöz; Halil İbrahim Bülbül

Makale detayları

Prediction of Higher Education Institutions Examination (YKS) Scores with Artificial Intelligence Techniques

Yazarlar:

Betül Ersöz | Halil İbrahim Bülbül

Yayıncı:

Gazi University

Ana Başlıklar:

Bilgi Modelleme | Yönetim ve Ontolojiler | Bilgi Sistemleri Eğitimi | Bilgi Sistemleri Kullanıcı Deneyimi Tasarımı ve Geliştirme

Dergi:

Gazi Üniversitesi Fen Bilimleri Dergisi Part C: Tasarım ve Teknoloji Vol.12 Issue 4, pp.845-853

Belge Tipi:

Makale

Yayın Tarihi:

12/2024

Belge Dili:

Türkçe

Document ID (Org No.):

2001337745

pISSN:

2147-9526

Anahtar kelimeler:

Özet:

 Students academic success and preferences primarily determine placements in higher education programs. Students who take the Higher Education Institutions Exam YKS are placed in higher education departments suitable for their career goals, primarily according to their YKS score, Secondary Education Success Score OBP, and preference rankings. The most crucial factor in placement with YKS is the exam score. Therefore, students need recommendation systems to determine their net scores before the exam. Similar to this study, various systems calculate students YKS scores using formulas from net scores of practice exams. However, it has been observed that no application estimates score with artificial intelligence methods. In this study, the YKS scores of students who participated to the YKS are estimated according to their net scores of practice exams. In the study, four different models such as Linear Regression LinReg, Multilayer Perceptron MLP, K-Nearest Neighbor KNN, Random Forest RF from Machine Learning ML algorithms, which are among the Artificial Intelligence AI techniques, were used in estimating the scores of TYT Basic Proficiency Test and AYT Field Proficiency Test in the two-stage YKS exam. Among the ML algorithms, the best-performing MLP for TYT was found to be R-Square 0.999, MAE 0.056, and RMSE 0.447. The best-performing Linear Regression for AYT was found to be R-Square 0.999, MAE 0.214, and RMSE 0.0413. The developed model is integrated into the generative artificial intelligence tool of a commercial digital learning platform by using ML techniques to predict students YKS score results and provides assistance in score predictions.

 Students academic success and preferences primarily determine placements in higher education programs. Students who take the Higher Education Institutions Exam YKS are placed in higher education departments suitable for their career goals, primarily according to their YKS score, Secondary Education Success Score OBP, and preference rankings. The most crucial factor in placement with YKS is the exam score. Therefore, students need recommendation systems to determine their net scores before the exam. Similar to this study, various systems calculate students YKS scores using formulas from net scores of practice exams. However, it has been observed that no application estimates score with artificial intelligence methods. In this study, the YKS scores of students who participated to the YKS are estimated according to their net scores of practice exams. In the study, four different models such as Linear Regression LinReg, Multilayer Perceptron MLP, K-Nearest Neighbor KNN, Random Forest RF from Machine Learning ML algorithms, which are among the Artificial Intelligence AI techniques, were used in estimating the scores of TYT Basic Proficiency Test and AYT Field Proficiency Test in the two-stage YKS exam. Among the ML algorithms, the best-performing MLP for TYT was found to be R-Square 0.999, MAE 0.056, and RMSE 0.447. The best-performing Linear Regression for AYT was found to be R-Square 0.999, MAE 0.214, and RMSE 0.0413. The developed model is integrated into the generative artificial intelligence tool of a commercial digital learning platform by using ML techniques to predict students YKS score results and provides assistance in score predictions.

 Students academic success and preferences primarily determine placements in higher education programs. Students who take the Higher Education Institutions Exam YKS are placed in higher education departments suitable for their career goals, primarily according to their YKS score, Secondary Education Success Score OBP, and preference rankings. The most crucial factor in placement with YKS is the exam score. Therefore, students need recommendation systems to determine their net scores before the exam. Similar to this study, various systems calculate students YKS scores using formulas from net scores of practice exams. However, it has been observed that no application estimates score with artificial intelligence methods. In this study, the YKS scores of students who participated to the YKS are estimated according to their net scores of practice exams. In the study, four different models such as Linear Regression LinReg, Multilayer Perceptron MLP, K-Nearest Neighbor KNN, Random Forest RF from Machine Learning ML algorithms, which are among the Artificial Intelligence AI techniques, were used in estimating the scores of TYT Basic Proficiency Test and AYT Field Proficiency Test in the two-stage YKS exam. Among the ML algorithms, the best-performing MLP for TYT was found to be R-Square 0.999, MAE 0.056, and RMSE 0.447. The best-performing Linear Regression for AYT was found to be R-Square 0.999, MAE 0.214, and RMSE 0.0413. The developed model is integrated into the generative artificial intelligence tool of a commercial digital learning platform by using ML techniques to predict students YKS score results and provides assistance in score predictions.

 Yükseköğretim programlarına yerleştirmeler öncelikle öğrencilerin akademik başarılarına ve tercihlerine göre belirlenir. Yükseköğretim Kurumları Sınavı YKS giren öğrenciler başta YKS puanı, Ortaöğretim Başarı Puanı OBP ve tercih sıralamalarına göre kariyer hedeflerine uygun yükseköğretim bölümlerine yerleştirilir. YKS ile yerleşmede en önemli faktör sınav puanıdır. Bu sebeple öğrenciler, sınav öncesi netlerinden puanlarının belirlenmesi için öneri sistemlerine ihtiyaç duymaktadır. Deneme sınavları netlerinden öğrencilerin YKS puanlarını formüller yardımıyla hesaplayarak tahmin eden hâlihazırda çeşitli sistemler mevcuttur. Ancak yapay zekâ yöntemleriyle puanları tahmin eden uygulamalar bulunmamaktadır. Bu çalışmada YKSye giren öğrencilerin deneme sınavı netlerine göre YKS puanlarının tahmini yapılmıştır. Araştırmada, iki aşamalı olan YKS sınavında TYT Temel Yeterlilik Testi ve AYT Alan Yeterlilik Testi puanlarının tahmin edilmesinde Yapay zekâ YZ tekniklerinden olan Makine Öğrenme ML algoritmalarından Lineer Regresyon LinReg, Çok Katmanlı Algılayıcı MLP, K-En Yakın Komşu KNN, Rastgele Orman RF gibi dört farklı model kullanılmıştır. ML algoritmaları içerisinde TYT de en iyi performans gösteren MLP, R-Kare 0.999, MAE 0.056 ve RMSE 0.447 değerleri bulunmuştur. AYT de en iyi performans gösteren Lineer regresyon R-Kare 0.999, MAE 0.214 ve RMSE 0.0413 değerleri bulunmuştur. Geliştirilen model, öğrencilerin YKS puan sonuçlarını ML teknikleriyle tespit ederek ticari bir dijital öğrenme platformun üretken yapay zeka aracına entegre edilmiş olup, puan tahminlerinde destek sağlamaktadır.

Makale detayları

Prediction of Higher Education Institutions Examination (YKS) Scores with Artificial Intelligence Techniques

Yazarlar:

Betül Ersöz | Halil İbrahim Bülbül

Yayıncı:

Gazi University

Ana Başlıklar:

Bilgi Modelleme | Yönetim ve Ontolojiler | Bilgi Sistemleri Eğitimi | Bilgi Sistemleri Kullanıcı Deneyimi Tasarımı ve Geliştirme

Dergi:

Gazi Üniversitesi Fen Bilimleri Dergisi Part C: Tasarım ve Teknoloji Vol.12 Issue 4, pp.845-853

Belge Tipi:

Makale

Yayın Tarihi:

12/2024

Belge Dili:

Türkçe

Document ID (Org No.):

2001337745

pISSN:

2147-9526

artificial intelligence

machine learning

YKS

TYT

AYT

Generative artificial intelligenceyapay zekâ

makine öğrenme

YKS

TYT

AYT

Üretken yapay zeka

Sayı

Makale detayları

Yazarlar: Betül Ersöz | Halil İbrahim Bülbül

Yayıncı: Gazi University

Ana Başlıklar: Bilgi Modelleme | Yönetim ve Ontolojiler | Bilgi Sistemleri Eğitimi | Bilgi Sistemleri Kullanıcı Deneyimi Tasarımı ve Geliştirme

Dergi: Gazi Üniversitesi Fen Bilimleri Dergisi Part C: Tasarım ve Teknoloji Vol.12 Issue 4, pp.845-853

Belge Tipi: Makale

Yayın Tarihi: 12/2024

Belge Dili: Türkçe

Document ID (Org No.): 2001337745

pISSN: 2147-9526

Sayı

Yazarlar:

Betül Ersöz | Halil İbrahim Bülbül

Yayıncı:

Gazi University

Ana Başlıklar:

Bilgi Modelleme | Yönetim ve Ontolojiler | Bilgi Sistemleri Eğitimi | Bilgi Sistemleri Kullanıcı Deneyimi Tasarımı ve Geliştirme

Dergi:

Gazi Üniversitesi Fen Bilimleri Dergisi Part C: Tasarım ve Teknoloji Vol.12 Issue 4, pp.845-853

Belge Tipi:

Makale

Yayın Tarihi:

12/2024

Belge Dili:

Türkçe

Document ID (Org No.):

2001337745

pISSN:

2147-9526