Comparing nine machine learning classifiers for school-dropouts using a revised performance measure

Sahar Saeed Rezk () and Kamal Samy Selim ()
Additional contact information
Sahar Saeed Rezk: Cairo University
Kamal Samy Selim: Cairo University

Journal of Computational Social Science, 2024, vol. 7, issue 2, No 16, 1555-1597

Abstract: Abstract Addressing the pervasive issue of school-dropout in Egypt is imperative for advancing the country's educational system and fostering its social and economic progress. Recently, there is a growing interest in leveraging Machine Learning techniques as proactive tools for identifying students at-risk of dropping out so as to carry out timely interventions. This study implements nine supervised Machine Learning algorithms, namely Decision Trees, K-Nearest Neighbours, Logistic Regression, Naïve Bayes, Support Vector Machines, AdaBoost, Bagging, Random Forest, and Stacking, and compares their results to figure out the best performing one for classifying at-risk students in the Egyptian compulsory schools. Utilizing a dataset of a nationally representative sample survey, 52 meticulous classification experiments combining classifiers and resampling techniques are conducted. For the classifiers admitting hyper-parameter optimization, 32 initial parameter settings entailing parameter-space searches, using GridSearch heuristic algorithm, are tried to determine the best performing configuration models for classification. Rather than relying on disparate performance measures for comparing the resulting classifications, such as accuracy and F-score, this research proposes the weighted harmonic mean of several performance measures as a unified evaluation criterion. By resorting to this single criterion for comparisons, the Support Vector Machines classifier, conjoint with Random Under-Sampling and Synthetic Minority Over-sampling Technique for treating class imbalance, is evaluated as the best performing classification model. Because of its ability to provide classification rules in explicit functional forms, Support Vector Machines enables interpreting the embedded features in a similar way like the Logistic Regression classifier. Consequently, the best results reached could guide to develop an early predicting system aiming to support the efforts to eradicate the persisting problem of school-dropouts in Egypt over time.

Keywords: Education; School-dropouts; Supervised learning; Intelligent prediction; Class imbalance; Performance measure (search for similar items in EconPapers)
Date: 2024
References: View references in EconPapers View complete reference list from CitEc
Citations: View citations in EconPapers (1)

Downloads: (external link)
http://link.springer.com/10.1007/s42001-024-00281-8 Abstract (text/html)
Access to the full text of the articles in this series is restricted.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:spr:jcsosc:v:7:y:2024:i:2:d:10.1007_s42001-024-00281-8

Ordering information: This journal article can be ordered from
http://www.springer. ... iences/journal/42001

DOI: 10.1007/s42001-024-00281-8

Access Statistics for this article

Journal of Computational Social Science is currently edited by Takashi Kamihigashi

More articles in Journal of Computational Social Science from Springer
Bibliographic data for series maintained by Sonal Shukla () and Springer Nature Abstracting and Indexing ().