Examining students’ attitudes towards STEM (science, technology, engineering, and mathematics) fields starting from middle school level is important in their career choices and future planning. However, there is a need to investigate which variables affect students’ attitudes towards STEM. Here, we aimed to estimate middle school students’ attitudes towards STEM with data mining algorithms using classification analysis. Data mining is one of the data analysis methods used successfully in different fields, including education, in recent years. 600 middle school students from different grade levels selected from various districts of Istanbul province participated in the study. The data collection tools of the research are the STEM Attitude Scale and Personal Information Form. The data obtained from the Personal Information Form is about the students’ school type, grade level, gender, academic achievement, mother and father occupation, education level of father and mother. According to the results of the research, the K-Star algorithm from the lazy group and the Random Tree algorithm from the trees group performed the best results in classifying data. According to the decision tree technique, the dominant factor influencing middle school students’ attitudes towards STEM fields is the grade levels. Besides, the factors that the K-Star algorithm finds important after grade level variable in classification are mother occupation and academic achievement level. It is hoped that this study will enlighten future research on setting an example for the use of data mining methods in educational research and determining the factors that affect students’ attitudes towards STEM fields.

从中学阶段开始检查学生对 STEM（科学、技术、工程和数学）领域的态度对于他们的职业选择和未来规划非常重要。然而，有必要调查哪些变量会影响学生对 STEM 的态度。在这里，我们旨在通过使用分类分析的数据挖掘算法来估计中学生对 STEM 的态度。数据挖掘是近年来成功应用于教育等不同领域的数据分析方法之一。来自伊斯坦布尔省各区的600名不同年级的中学生参加了研究。该研究的数据收集工具是 STEM 态度量表和个人信息表。从个人信息表中获取的数据包括学生的学校类型、年级、性别、学业成绩、父母职业、父母教育程度。根据研究结果，lazy 组的 K-Star 算法和 trees 组的 Random Tree 算法在数据分类方面表现最好。根据决策树技术，影响中学生对STEM领域态度的主导因素是年级水平。此外，K-Star 算法发现在分类中的年级水平变量之后的重要因素是母亲职业和学业成绩水平。