AI 피드백을 활용한 중학생의 통계적 탐구 과정 분석: 웹 로그 데이터 기반 학습 분석 관점을 중심으로 = Analyzing middle school students’ statistical inquiry processes using AI-generated feedback: A learning analytics approach based on web log data

다국어 초록 (Multilingual Abstract)

This study investigates how middle school students utilize generative AI feedback in a web-based statistical inquiry environment and examines changes in their inquiry strategies and affective domains using log data-based learning analytics. Using a GPT-4o API-enhanced tool, the inquiry processes of 13 students were collected. K-means clustering identified three inquiry patterns: ‘Visualization-oriented’, ‘Analysis-oriented’, and ‘Balanced’. The Analysis-oriented group improved consistently across stages, while the Balanced group excelled in problem definition. Notably, the Visualization-oriented group showed significant growth in interpretation through graph exploration. Post-surveys indicated increased interest and value. Specifically, the Analysis-oriented group showed increased value perception, the Balanced group improved in confidence, and the Visualization-oriented group showed increased interest. These findings suggest that the effects of AI feedback vary by inquiry style and highlight the utility of log-based diagnostics for designing adaptive support.