Bilimsel süreç becerilerinin yapay zekâ ile yordanması, öğrenciler ve üstün yetenekli öğrencilerdeki etkililiği
Loading...
Date
2023-09-14
Authors
Sarıoğlu, Serhan
Journal Title
Journal ISSN
Volume Title
Publisher
Bursa Uludağ Üniversitesi
Abstract
Bilimsel süreç becerileri, bilimsel bir araştırmayı yürütme sürecinde kullanılması gereken ve bu nedenle öğrencilere kazandırılması gereken en önemli becerilerin başında gelmektedir. Bilimsel süreç becerileri fen bilimlerinin öğretiminde ve bilimle ilgili tüm mesleklerin edinilmesi sürecinde önemli bir rol oynamaktadır. Özellikle Türk eğitim sisteminde sıklıkla kullanılan çoktan seçmeli sorular, bilimsel süreç becerileri gibi işe vuruk becerilere odaklanmaması, öğretim programının vurguladığı kazanımlarla sınırlı kalması nedeniyle bu becerilerin ölçülmesini ve değerlendirilmesini, dolayısıyla da bu noktadaki eksikliklerin ortaya çıkarılarak bu eksikliklere yönelik önlemler almak suretiyle geliştirilmesini zorlaştırmaktadır. Bu nedenle, araştırma “Geliştirilen sınıflama modelinin beceri temelli sorularla 8. Sınıf öğrencilerinin ve fen alanında üstün yetenekli 8. sınıf öğrencilerinin bilimsel süreç becerilerini yordamada etkililiği nasıldır?” sorusuna yanıt aramaktadır. Araştırmada kullanılan veri toplama araçları çevrimiçi bilimsel süreç becerileri testi ve40 sorudan oluşan çoktan seçmeli beceri temelli sorulardan oluşmaktadır. Veri toplama araçlarının uygunluğu, ölçmeyi hedeflediği bilimsel süreç becerilerinin standartlarıyla uyumsama durumları gözetilerek geliştirilmiş ve öğrencilerin bilimsel süreç beceri düzeylerini ortaya çıkarmak amacıyla kullanılan bilimsel süreç beceri ölçeği ile birlikte araştırmanın veri toplama araçlarını oluşturmuştur. Bu aşamadan sonra 292 öğrenciden elde edilen veriler yardımıyla beceri temelli maddelerin bilimsel süreç becerilerini yordama durumlarını ortaya çıkarmak için 11 farklı makine öğrenmesi sınıflama modeli oluşturulmuş, bu modellerin etkililikleri yine nicel metriklerden yararlanılarak değerlendirilmiştir. Eğitilen sınıflama modeli daha sonra 34 fen alanında üstün yetenekli tanısı almış öğrencilerin verileri üzerinde test edilmiş ve sonuçları raporlanmıştır. Araştırma sonucunda geliştirilen farklı modeller değerlendirilmiş ve öğrencilerin bilimsel süreç becerilerini yordamada en iyi performansı gösteren modelin Adaboost modeli olduğu ortaya çıkarılmıştır. Ayrıca modelin performansı üstün yetenekli öğrencilerin bilimsel süreç becerilerinin yordanmasında da incelenmiş, sonuçta kabul edilebilir düzeyde bir başarı ortaya çıkarılmıştır. Araştırma sonucunda yapay zekâ uygulamalarının eğitimde ölçme değerlendirme alanında, özellikle de bilimsel süreç becerileri gibi özel bir alanda bile kullanılabileceği sonucuna ulaşılmış, daha farklı örneklemlerle farklı öğrenci özelliklerinin de yordanması için daha fazla çalışmanın yapılması önerilmiştir. Bununla birlikte, sınıflama modelinin eğitilmesi sürecinde kullanılan veri noktası sayısının mümkün olduğunca fazla olmasının modelin performansını olumlu yönde etkileyeceği önerilmektedir.
Science process skills are one of the most important skills to be used in the process of conducting scientific research and should therefore be acquired by students. Science process skills play an important role in the teaching of science and in the acquisition of all science-related professions. Multiple-choice questions, which are frequently used in the Turkish education system, do not focus on hands-on skills such as science process skills and are limited to the outcomes emphasised by the curriculum, which makes it difficult to measure and evaluate these skills and therefore to identify the deficiencies at this point and to improve them by taking measures to address these deficiencies. Therefore, the research seeks to answer the question "What is the effectiveness of the developed classification model in predicting the science process skills of students and gifted students with skill-based questions? The data collection tools used in the study consisted of an online science process skills test and 40 multiple-choice skill-based questions. The data collection instruments were developed considering the appropriateness of the data collection instruments, their compatibility with the standards of the scientific process skills that were to be measured, and together with the Science Process Skills Scale, which was used to reveal the level of students' scientific process skills, constituted the data collection instruments of the study. After this phase, using the data obtained from 292 students, 11 different machine-learning classification models were developed to reveal the predictive status of items based on scientific process skills, and the effectiveness of these models was evaluated using quantitative measures. The trained classification model was then tested on data from 34 students diagnosed as gifted in science and the results reported. As a result of the research, different models were evaluated, and it was found that the Adaboost model was the best performing model in predicting students' science process skills. In addition, the performance of the model in predicting the science process skills of gifted students was also investigated and an acceptable level of success was found. As a result of the research, it was concluded that artificial intelligence applications can be used in the field of measurement and evaluation in education, especially in a specialised area such as science process skills, and it is recommended that further studies should be carried out to predict different student characteristics with different samples. In addition, it is suggested that the number of data points used in the training process of the classification model should be as large as possible, as this will have a positive effect on the performance of the model.
Science process skills are one of the most important skills to be used in the process of conducting scientific research and should therefore be acquired by students. Science process skills play an important role in the teaching of science and in the acquisition of all science-related professions. Multiple-choice questions, which are frequently used in the Turkish education system, do not focus on hands-on skills such as science process skills and are limited to the outcomes emphasised by the curriculum, which makes it difficult to measure and evaluate these skills and therefore to identify the deficiencies at this point and to improve them by taking measures to address these deficiencies. Therefore, the research seeks to answer the question "What is the effectiveness of the developed classification model in predicting the science process skills of students and gifted students with skill-based questions? The data collection tools used in the study consisted of an online science process skills test and 40 multiple-choice skill-based questions. The data collection instruments were developed considering the appropriateness of the data collection instruments, their compatibility with the standards of the scientific process skills that were to be measured, and together with the Science Process Skills Scale, which was used to reveal the level of students' scientific process skills, constituted the data collection instruments of the study. After this phase, using the data obtained from 292 students, 11 different machine-learning classification models were developed to reveal the predictive status of items based on scientific process skills, and the effectiveness of these models was evaluated using quantitative measures. The trained classification model was then tested on data from 34 students diagnosed as gifted in science and the results reported. As a result of the research, different models were evaluated, and it was found that the Adaboost model was the best performing model in predicting students' science process skills. In addition, the performance of the model in predicting the science process skills of gifted students was also investigated and an acceptable level of success was found. As a result of the research, it was concluded that artificial intelligence applications can be used in the field of measurement and evaluation in education, especially in a specialised area such as science process skills, and it is recommended that further studies should be carried out to predict different student characteristics with different samples. In addition, it is suggested that the number of data points used in the training process of the classification model should be as large as possible, as this will have a positive effect on the performance of the model.
Description
Keywords
Adaboost, Bilimsel süreç becerileri, Makine öğrenmesi, Üstün yetenekliler, Yapay zekâ, Artificial intelligence, Gifted students, Machine learning, Science process skills
Citation
Sarıoğlu, S. (2023). Bilimsel süreç becerilerinin yapay zekâ ile yordanması, öğrenciler ve üstün yetenekli öğrencilerdeki etkililiği.Yayınlanmamış doktora tezi. Bursa Uludağ Üniversitesi Eğitim Bilimleri Enstitüsü.