ACADEMIC BEHAVIOR PATTERN MINING USING SPADE ALGORITHM IN EDUCATIONAL TRAJECTORY DISCOVERY SYSTEMS
Roberto Acevedo
Keywords: Academic, Behavior, Pattern, Mining, Sequential, Pattern, Discovery, Equivalence, Educational, Trajectory, Discovery Systems
Abstract
Studying academic behavior patterns is essential for predicting student performance and for devising effective learning interventions. Traditional data mining techniques often do not account for the sequence of academic events and the sequential nature of learning, and therefore are limited in the capacity to discover learning trajectories that make sense. In this paper, an Educational Trajectory Discovery System (ETDS) in education to analyse student academic behavior and utilize the SPADE (Sequential Pattern Discovery using Equivalence classes) algorithm has been proposed for identifying frequent sequential patterns in student activities. The system works with various academic data sources (e.g., learning management system logs, assessments, and attendance and participation data) and organizes the data into event sequences. SPADE finds sequential behavior patterns effectively using vertical id-list databases and equivalence class discovery, which reduces the computational cost by maximizing the benefit of the temporal relationship of the data while retaining its meaning. Event patterns extracted will then be mapped to academic performance outcomes of the learner and categorized into risk-prone, average, or high-achieving academic pathways. This will enable an early assessment of a learner at risk. A visual analytics dashboard improves the interpretability through trajectory visualizations, sequential rule graphs, and trend evolution across semesters. Educators gain actionable intelligence to design specific interventions while learners earn personalized recommendations to improve their learning strategies. Based on simulation results, the proposed system outperforms the standard sequence mining methods in discovery effectiveness, accuracy, and ability to identify at-risk students sooner. The results establish the proposed system as a robust, scalable, data-driven approach to academic monitoring, enabling proactive decision-making and encouraging tailored student learning paths within academic contexts.