Identification and Analysis of Blockly Programming Learning Behavior Patterns among Middle School Students Based on Learning Behavior Data

Abstract

This study examines structural differences in learning
behaviors among middle school students in block-based
programming. Using data from 55 students completing the
same task on an AI-PBL platform, raw event logs were
recoded into six categories of functional behaviors, and
student-level feature vectors were constructed. Clustering
analysis was applied to group students and identify three
behavior patterns: Exploration–Debugging,
Construction–Adjustment, and Monitoring-Oriented. These
patterns differ in interaction intensity and behavioral
structure. From the perspective of behavioral structure,
the study provides an analytical approach for examining
differences in students’ programming behavior patterns
based on learning behavior data.