Predicting Students Performance By Analyzing Their Behavior In E-Learning Systems Through Interaction Pattern Mining

Abstract

This study focuses on interaction pattern mining within students' learning trajectories and addresses the impact of student interaction patterns on their performance in e-learning courses. Typically, instructors structure the course sequence based on their didactic and pedagogical strategies, with the intention of guiding students through their learning journey. However, in the absence of strict constraints, students might opt for learning paths that diverge from the predefined sequence. This context prompts an important question: What are the consequences for student learning outcomes when they pursue learning paths that deviate from the instructor's expectations? Within E-learning platform, students' interactions with course materials are logged as events. Employing Educational Process Mining techniques allows for the extracting statistical information, tracing and modeling of student actions during and Sequential Pattern Mining (SPM) used for sequential patterns within learning process.. We develop an LMS with which students can interact in both a directed and free manner. We utilized an event log containing 37,405 events, gathered from 76 undergraduate students. Prior to analysis, this log underwent a preprocessing phase. For experiment, we segmented log data into three distinct datasets. To derive statistical insights, we employed the PROM framework. Our investigation entailed the application of four distinct process discovery algorithms namely, Alpha Miner, Heuristic Miner, ILP Miner, and Inductive Miner also GSP algorithm were implemented through scripts based on the PM4PY library. The outcomes of our study revealed that students exhibited unique behaviors while accessing the LMS and engaging in activities. Interestingly, we observed that students who followed a predetermined sequence or interacted with the LMS in a guided manner achieved higher grades compared to their counterparts who navigated the LMS in a more random fashion.