EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

An Information Technology System for Measuring Organizational Culture and Predicting Employee Turnover Based on Questionnaire Data and Text Mining

Firas Layth Khaleel Alameen, Farah Amer Abdalaziz, Mohammed Abdulaziz Alsubh, Abdullatif Saleh Alfaqiri, Normala Rahim, Mohammed Nizam Saad, Tengku Siti Meriam Tengku Wook

Wasit Journal of Computer and Mathematics Science · 30.06.2026 · 5 (2) · 51-72

DOI: 10.31185/wjcms.522 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Wasit Journal of Computer and Mathematics Science, online erschienen am 30.06.2026. Verfasst von Firas Layth Khaleel Alameen und 6 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This paper implements the prediction of employees leaving their company using conventional methods such as surveys and modern techniques like mining textual data. The open-ended responses were analyzed (12,847 words of text in total) resulting in a vocabulary of 1,892 unique words. The words that came up most often in positive reference were “team” (142), “growth” (98) and “resilience” (87); negative terms found included “pressure”, which occurred 156 times, along with injustice (112) and burnout (76). The average sentiment score was 2.8/5, with 34% of responses very negative and only 18% very positive. Various machine learning models have been tested to predict for turnover. If we only use closed-ended questions, though, XGBoost reaches an accuracy score of 78%. But then features for text mining (sentiment scores + keywords) increased accuracy to 87% (+9%), precision:0.85, recall:0.84, AUC-ROC 0.89 Key risk indicators based on text mining included the term “unfair” (turnover risk 4.2x higher), “burnout” (3.8x higher risk) while “recognition” (3.5x lower risk) and growth (2.8 times lower risk). The turnover risk increased by 31% per 1-point decrease in sentiment score. The hybrid XGBoost model with text features performed better than fine-tuned BERT ( accuracy of 84% ). This proposed IT stains enables organization to proactively identify risk oriented employees through linguistic signals in the open-ended feedback allowing timely intervention by HR wherein it improves efficiency.

Zitieren

APA

Alameen, F. L. K., Abdalaziz, F. A., Alsubh, M. A., Alfaqiri, A. S., Rahim, N., Saad, M. N., & Wook, T. S. M. T. (2026). An Information Technology System for Measuring Organizational Culture and Predicting Employee Turnover Based on Questionnaire Data and Text Mining. Wasit Journal of Computer and Mathematics Science, 5(2), 51-72. https://doi.org/10.31185/wjcms.522

Vancouver

Alameen FLK, Abdalaziz FA, Alsubh MA, Alfaqiri AS, Rahim N, Saad MN, et al. An Information Technology System for Measuring Organizational Culture and Predicting Employee Turnover Based on Questionnaire Data and Text Mining. Wasit Journal of Computer and Mathematics Science. 2026;5(2):51-72. doi:10.31185/wjcms.522

BibTeX
@article{Alameen2026,
  author = {Alameen, Firas Layth Khaleel and Abdalaziz, Farah Amer and Alsubh, Mohammed Abdulaziz and Alfaqiri, Abdullatif Saleh and Rahim, Normala and Saad, Mohammed Nizam and Wook, Tengku Siti Meriam Tengku},
  title = {An Information Technology System for Measuring Organizational Culture and Predicting Employee Turnover Based on Questionnaire Data and Text Mining},
  journal = {Wasit Journal of Computer and Mathematics Science},
  year = {2026},
  volume = {5},
  number = {2},
  pages = {51--72},
  doi = {10.31185/wjcms.522}
}
Kostenlos starten →