ClickTree: A Tree-based Method for Predicting Math Students’ Performance Based on Clickstream Data
Core Concepts
ClickTree is a tree-based methodology developed to predict student performance in mathematical assignments using clickstream data, achieving an AUC of 0.78844 and ranking second in the Educational Data Mining Cup 2023.
Abstract
1. Introduction:
Analyzing clickstream data for insights into student behavior and academic outcomes.
Importance of predicting student performance in online courses.
2. Material and Method:
Description of the dataset from EDM Cup 2023.
Feature extraction process including assignment-level, student-level, and problem-level features.
Utilization of CatBoost classifier for performance prediction.
3. Results:
Identification of challenging problem types and subjects for students.
Comparison between successful and struggling students' learning behaviors.
Evaluation of ClickTree method with an AUC of 0.78844.
4. Implications and Recommendations:
Suggestions to improve teaching strategies based on identified challenges.
Providing targeted support to struggling students based on their interactions with in-unit assignments.
5. Conclusion and Future Work:
Development of ClickTree method for predicting student performance.
Importance of validation set selection for model accuracy assessment.
Customize Summary
Rewrite with AI
Generate Citations
Translate Source
To Another Language
Generate MindMap
from source content
Visit Source
arxiv.org
ClickTree
Stats
学生のパフォーマンスを予測するために開発されたClickTreeは、Educational Data Mining Cup 2023でAUCが0.78844で2位にランクインしました。