Article Details

Abstract

With the popularization and advancement of higher education, an increasing number of individuals with professional expertise and innovative capabilities are entering society, thereby promoting social progress and development. Meanwhile, the issue of college student dropout has attracted growing attention. Traditional Machine Learning methods suffer from limited interpretability in dropout prediction. To address this problem, this paper constructs a scorecard model for college dropout early warning. Firstly, student data are cleaned and preprocessed. Feature binning and WOE (Weight of Evidence) encoding are then performed, followed by model construction using L1-regularized logistic regression. Finally, the model is transformed into a scorecard. Model evaluation shows that the model achieves an AUC of 0.949, demonstrating strong predictive power and interpretability. The model can provide a reference for college dropout early warning and early targeted intervention.

Keywords

scorecard model; college student dropout; L1 regularization; Logistic Regression

References

1. [1] Ministry of Education of the People's Republic of China. Statistical Bulletin on the Development of National Education in 2019 [EB/OL]. (2020-05-20) [225-11-28] http://www.moe.gov.cn/jyb_sjzl/sjzl_fztjgb/202005/t20200520_456751.html.
2. [2] Ministry of Education of the People's Republic of China. Statistical Bulletin on the Development of National Education in 2024 [EB/OL]. (June 11, 2025) [November 28, 2025] http://www.moe.gov.cn/jyb_sjzl/sjzl_fztjgb/202506/t20250611_1193760.html.
3. [3] Bettahi A，Belouadha F-Z，Harroud H. A Modular and Explainable Machine Learning Pipeline for Student Dropout Prediction in Higher Education [J/OL].Algorithms，2025，18（10）：662.https://doi.org/10.3390/a18100662.
4. [4] Tang Minghao Analysis of Reasons and Early Warning for College Students Dropout Based on Optimizing XGBoost HAP [D]. Dalian: Northeastern University of Finance and Economics, 2024
5. [5] Agrusti F，Mezzini M，Bonavolontà G. Deep Learning Approach for Predicting University Dropout: A Case Study at Roma Tre University [J].Journal of E-Learning and Knowledge Society，2020，16（1）：44-54.
6. [6] Cho C H，Yu Y W，Kim H G. A Study on Dropout Prediction for University Students Using Machine Learning [J/OL].Applied Sciences，2023，13（21）：12004.https://doi.org/10.3390/app132112004.
7. [7] Marcolino M R，Porto T R，Primo T T，et al. Student Dropout Prediction through Machine Learning Optimization: Insights from Moodle Log Data [J/OL].Scientific Reports，2025，15（1）：9840.https://www.nature.com/articles/s41598-025-93918-1.
8. [8] Yan Xu Risk assessment model for recidivism of released prisoners based on credit score cards [D]. Changchun: Jilin University, 2022
9. [9] Liu Zidi Research and Application of Medical Health Risk Assessment Model Based on AI Technology [D]. Urumqi: Xinjiang University, 2023
10. [10] Mi Yanbiao Research on Driving Behavior Safety Scoring Method Based on Connected Car Data [D]. Changchun: Jilin University, 2023
11. [11] Martins M V，Tolledo D，Machado J，et al. Early Prediction of Student's Performance in Higher Education: A Case Study [C]//World Conference on Information Systems and Technologies，2021：166-175.
12. [12] Rong Yixuan Research on the Prediction Method of Student Academic Performance Based on Deep Learning [D]. Lanzhou: Northwest Normal University, 2024
13. [13] Yang Zhou, Cui Caixia Early warning of academic risks among college students based on heterogeneous Stacking ensemble learning [J]. Journal of Taiyuan Normal University (Natural Science Edition), 2025, 24 (1): 22-29, 39
14. [14] Zhang Junli, Guo Shuangyan, Ren Cuiping, etc Research on Personal Credit Scoring Card Model Based on Logistic Regression [J]. Modern Information Technology, 2024, 8 (5): 12-16
15. [15] Niu Zhihui, Chen Bo, Bu Chunyao Prediction of velocity large pulses and analysis of influencing factors based on L1 regularized logistic regression model [J]. Journal of Earthquake Engineering, 2022, 44 (2): 306-320
16. [16] Song Z H，Sung S-H，Park D-M，et al. All-Year Dropout Prediction Modeling and Analysis for University Students [J/OL].Applied Sciences，2023，13（2）：1143.https://doi.org/10.3390/app13021143.

Funding

Natural Science Foundation of Jilin Province - Key Research and Development (20240304097SF); Jilin Provincial Natural Science Foundation - Basic Research (YDZJ202301ZYTS157)

About the authors

Zhou Zhihan (2002-), female, Han ethnicity, from Liaoyuan, Jilin Province, is currently pursuing a master's degree in machine learning and educational data mining.