Predicting outcomes in early childhood education using a neutrosophic Random Forest model: Managing uncertainty in pedagogical decision-making

Abstract

This article presents a novel framework that integrates Random Forest classification with neutrosophic logic to address the challenge of uncertainty-aware decision-making in nursery school application processes. Using the publicly available Nursery dataset, which includes socio-familial attributes such as parental occupation, financial status, and health conditions, the proposed model not only achieves high predictive accuracy (approximately 95%) but also explicitly quantifies uncertainty through neutrosophic sets defined by degrees of truthfulness (V), indeterminacy (I), and falsity (F). This approach enables a nuanced interpretation of classification confidence, distinguishing between clear cases that can be automated and borderline instances that require human expert review. By enabling a transparent and tiered decision-making strategy, the framework improves the fairness, explainability, and operational efficiency of admissions systems, offering a practical tool for administrative use in high-risk educational settings.