Santoso Purwoko Haryadi, Setiaji Bayu, Kurniawan Yohanes, Bahri Syamsul, Kusuma Mobinta, Wusqo Indah Urwatin, Muldayanti Nuri Dewi, Kurniawan Arif Didik, Syahbrudin Johan
Department of Physics Education, Universitas Sulawesi Barat, Majene, 91413, Indonesia.
Department of Physics Education, Universitas Negeri Yogyakarta, Sleman, 55281, Indonesia.
Sci Data. 2025 Jun 12;12(1):987. doi: 10.1038/s41597-025-04913-0.
There is a need to help advance research on using machine learning and data mining techniques in physics education research (PER), which might still be difficult due to the unavailable dataset for the specific purpose of PER. The SPHERE (Students' Performance Dataset in Physics Education Research) is presented as an educational dataset of physics learning collected through research-based assessments (RBAs) established by the PER scholars. In this study, students' performance in physics at four public high schools was probed in three learning domains. It encompassed students' conceptual understanding, scientific ability, and learning attitude toward physics. The employed RBAs were identified based on the curriculum of physics contents taught to the eleventh-grade students in the ongoing academic year. In this paper, we provide an example that SPHERE could be insightful for training machine learning models to predict students' performance at the end of the learning process. We also revealed that its predictive performance was superior to the former method of students' performance prediction as labeled by the physics teachers.
有必要推动在物理教育研究(PER)中使用机器学习和数据挖掘技术的研究,由于缺乏用于PER特定目的的数据集,这可能仍然很困难。SPHERE(物理教育研究中的学生表现数据集)作为通过PER学者建立的基于研究的评估(RBA)收集的物理学习教育数据集被呈现出来。在本研究中,对四所公立高中学生在三个学习领域的物理表现进行了探究。它涵盖了学生的概念理解、科学能力以及对物理的学习态度。所采用的RBA是根据本学年向十一年级学生教授的物理内容课程确定的。在本文中,我们提供了一个例子,说明SPHERE对于训练机器学习模型以预测学习过程结束时学生的表现可能具有深刻见解。我们还揭示,其预测性能优于物理教师之前标记的学生表现预测方法。