EUVIMEDEUVIMED entdecken
Open Access · CC BY-SA 4.0

Classifying Job-Posting Wage Compliance Using Machine Learning: A Comparative Study of Random Forest and Support Vector Machine Algorithms

Tiffany Phylicia, Alyssa Christiana Lin, Glaudio Hiewen Tjongdro, Dylan Rael Andrew Bojoh, Evander Banjarnahor

Journal of Vocational, Informatics and Computer Education · 10.06.2026 · 234-245

DOI: 10.66053/voice.v4i2.649 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Vocational, Informatics and Computer Education, online erschienen am 10.06.2026. Verfasst von Tiffany Phylicia und 4 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-SA 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Purpose – This study examines wage compliance in online job postings by classifying posted salaries relative to the 2025 DKI Jakarta Provincial Minimum Wage (UMP). It aims to describe below-threshold salary postings and evaluate how salary-range variables affect machine-learning classification. Methods – A raw dataset of 1,221 job postings was collected from Loker.id on October 5, 2025. After removing duplicates, missing values, and salary outliers, the final dataset consisted of 1,143 postings. The target variable was constructed by comparing the mean posted salary range with the UMP threshold. Random Forest and Support Vector Machine (SVM) were evaluated under two scenarios: with and without salary-range variables. Performance was assessed using accuracy, balanced accuracy, F1-score, confusion matrices, and a majority-class baseline. Findings – The descriptive results show that 798 postings, or 69.82%, were classified as Below UMP Jakarta, while 345 postings, or 30.18%, were classified as Meets/Exceeds UMP Jakarta. With salary features included, Random Forest achieved 0.9446 test accuracy and SVM achieved 0.9592. Without salary features, performance declined to 0.7318 for Random Forest and 0.6968 for SVM, with the latter close to the majority-class baseline of 0.6982. Research implications – The findings suggest that the descriptive contribution of this study is stronger than its predictive contribution. Salary-range variables strongly influence classification performance because they are directly related to the construction of the target label. Therefore, machine-learning results should be interpreted cautiously and should not be treated as evidence of robust wage-compliance prediction from broader HR attributes alone. Originality – This study contributes to online labor-market analysis by combining descriptive wage-compliance evidence with an explicit feature-scenario comparison. By evaluating models with and without salary-range variables, the study highlights the importance of addressing threshold-related leakage in job-posting salary classification.

Zitieren

APA

Phylicia, T., Lin, A. C., Tjongdro, G. H., Bojoh, D. R. A., & Banjarnahor, E. (2026). Classifying Job-Posting Wage Compliance Using Machine Learning: A Comparative Study of Random Forest and Support Vector Machine Algorithms. Journal of Vocational, Informatics and Computer Education, 234-245. https://doi.org/10.66053/voice.v4i2.649

Vancouver

Phylicia T, Lin AC, Tjongdro GH, Bojoh DRA, Banjarnahor E. Classifying Job-Posting Wage Compliance Using Machine Learning: A Comparative Study of Random Forest and Support Vector Machine Algorithms. Journal of Vocational, Informatics and Computer Education. 2026:234-245. doi:10.66053/voice.v4i2.649

BibTeX
@article{Phylicia2026,
  author = {Phylicia, Tiffany and Lin, Alyssa Christiana and Tjongdro, Glaudio Hiewen and Bojoh, Dylan Rael Andrew and Banjarnahor, Evander},
  title = {Classifying Job-Posting Wage Compliance Using Machine Learning: A Comparative Study of Random Forest and Support Vector Machine Algorithms},
  journal = {Journal of Vocational, Informatics and Computer Education},
  year = {2026},
  pages = {234--245},
  doi = {10.66053/voice.v4i2.649}
}
Kostenlos starten →