Predictive Modeling and Explainability of Student Employability in the Philippines Using Random Forest and Shapley Additive Explanations
This study aims to analyze and predict the employability outcomes of higher education students in the Philippines using data-driven predictive modeling. It aims to identify the student attributes that most strongly influence employability, thereby supporting evidence-based curriculum development and workforce alignment initiatives.
Graduate employability remains a challenge in the Philippines due to persistent gaps between higher education outcomes and labor market expectations. While prior studies emphasize technical competencies, there is limited evidence on the role of non-academic student attributes in predicting employability, particularly through interpretable predictive approaches. This study addresses this gap by examining employability-related traits using machine learning techniques.
A publicly available dataset of 2,982 anonymized student records from mock interviews conducted across Philippine higher education institutions was analyzed. Employability was treated as a binary outcome variable. Predictive models were developed with Random Forest selected as the primary model based on overall performance. Model interpretation was conducted using Shapley Additive Explanations (SHAP) analysis to identify the most influential student attributes.
The study demonstrates how interpretable machine learning can be used to evaluate graduate employability and identify key attributes shaping workforce readiness. The findings offer practical value for higher education institutions and policymakers seeking data-driven approaches to curriculum design and student development.
The Random Forest model showed strong predictive performance in classifying employability outcomes. SHAP analysis revealed that mental alertness, general appearance, and the ability to present ideas were the most influential factors affecting employability, indicating the importance of cognitive and professional presentation skills.
Higher education institutions should strengthen instructional strategies and student development programs that enhance cognitive readiness, professional presentation, and communication-related competencies. Career services and industry partners are encouraged to collaborate in aligning training initiatives with employability needs.
Future studies should explore additional predictive models and incorporate broader datasets to further validate the determinants of employability. Including demographic and academic variables may deepen understanding of employability dynamics.
By identifying critical employability attributes, this study supports more responsive and inclusive higher education practices. Data-driven employability strategies can contribute to improved workforce readiness and reduced graduate unemployment in the Philippines.
Future research may examine longitudinal employability outcomes and assess the application of predictive analytics in institutional decision-making contexts. Attention to ethical considerations, including transparency and fairness, is also recommended.
Back