Date of Publication

4-15-2025

Document Type

Bachelor's Thesis

Degree Name

Bachelor of Science in Statistics Major in Actuarial Science

Subject Categories

Educational Psychology | Mathematics

College

College of Science

Department/Unit

Mathematics and Statistics Department

Thesis Advisor

Vio Jianu C. Mojica

Defense Panel Chair

Angelo M. Alberto

Defense Panel Member

Kevynn P. Delgado

Abstract/Summary

Bullying among adolescents is a common occurrence globally, negatively impacting victims’ self-esteem and well-being over time. 65% of Filipino students reported experiencing bullying at least a few times per month, the highest among participating countries according to the results of the 2018 Programme for International Student Assessment (PISA). Thus, this study aimed to identify factors associated with bullying victimization, providing evidence-based targeted interventions for reducing prevalence in the Philippines. Machine learning techniques, such as random forest, Naïve Bayes, and logistic regression, were applied to predict bullying victimization. The classification models were subsequently assessed through five-fold cross-validation with recall, precision, F1 score, and Matthews correlation coefficient computed as performance measures. Shapley Additive Explanations (SHAP) was used to interpret feature contributions to model predictions based on magnitude and direction. Key factors associated with bullying victimization included well-being, teaching methods, school dynamics, belonging, cultural respect and awareness, resilience, socioeconomic resources, technology access, and academic performance. Vulnerable groups included top-performing and underperforming students, those with a low sense of belonging at school, and those with poorly perceived disciplinary climates. Among the three classification algorithms, random forest achieved the best predictive performance and was consequently analyzed using SHAP for added interpretability. The SHAP analysis showed that students who experienced bullying victimization have low subjective well-being, low sense of belonging, and perceived lenient rules enforcement, strict teacher-directed instruction, perceived more competence and cooperation, and high respect and resiliency. The findings of this study may guide educators and governing bodies in developing targeted strategies to address bullying and foster safer, more inclusive school environments.

Abstract Format

html

Language

English

Format

Electronic

Keywords

Bullying in schools--Philippines; Victims of bullying--Philippines; Machine learning; Logistic regression analysis

Upload Full Text

wf_yes

Embargo Period

4-15-2025

Download

Share

COinS