Abstract: Youth violence and institutional safety incidents emerge from complex interactions among individual behaviour, social exposure, community conditions, prior institutional contact, and administrative records. Conventional prediction models often analyse these domains independently, limiting their capacity to represent nonlinear dependencies, temporal patterns, and cumulative risk pathways. This study proposes a novel Social-Behavioral-Community-Administrative Risk Fusion Algorithm (SBCA-RFA) for predicting youth violence and institutional safety outcomes through integrated multi-domain risk modelling. The proposed algorithm combines structured social indicators, behavioural histories, community-level vulnerability measures, and administrative risk variables within a weighted feature-fusion architecture incorporating gradient-based feature learning, attention-guided risk weighting, interaction modelling, and calibrated probabilistic classification. The system is designed to generate both individual violence-risk probabilities and institutional safety-risk scores while retaining interpretable contributions from major predictor groups. Comparative experiments evaluate SBCA-RFA against Logistic Regression, Support Vector Machine, Random Forest, Extreme Gradient Boosting, Light Gradient Boosting Machine, and a conventional Multilayer Perceptron. Performance is assessed using accuracy, precision, recall, F1-score, area under the Receiver Operating Characteristic curve, area under the Precision-Recall curve, Brier score, calibration error, false-negative rate, and balanced accuracy. Receiver Operating Characteristic curves, Precision-Recall curves, calibration plots, confusion matrices, feature-importance graphs, risk-distribution plots, and algorithm-comparison charts are used to examine predictive discrimination, reliability, interpretability, and operational usefulness. Ablation experiments further quantify the contribution of social, behavioural, community, and administrative indicator groups to predictive performance. Explainability is incorporated using Shapley Additive Explanations to identify the variables and cross-domain interactions most strongly associated with predicted violence and safety outcomes. The proposed framework is intended to demonstrate that integrated multi-source risk representation can provide stronger discrimination and more reliable early-warning capability than single-domain and conventional machine-learning approaches. The resulting system provides a technically rigorous foundation for data-informed prevention, targeted youth support, institutional resource allocation, and proactive safety management while emphasizing transparent and responsible risk assessment.
Keywords: Youth Violence; Institutional Safety; Social Indicators; Behavioral Risk; Administrative Risk.
Title: Predicting Youth Violence and Institutional Safety Outcomes through the Integration of Social Behavioral Community and Administrative Risk Indicators
Author: Godbless Amokwaw
International Journal of Management and Commerce Innovations
ISSN 2348-7585 (Online)
Vol. 14, Issue 1, April 2026 - September 2026
Page No: 749-773
Research Publish Journals
Website: www.researchpublish.com
Published Date: 18-August-2026
