• D. Shravani
• Jyothi Urade
• Akshaya Nandhyala
• Ms. M. Amitha
• Dr. S. JayaLaxmi

Recidivism, the tendency of released individuals to reoffend, poses a significant challenge to criminal justice systems worldwide and public safety. Traditional risk assessment tools are typically based on static factors and subjective measures, contributing to biased and inconsistent assessments. Our paper presents a machine learning (ML) model to predict whether an individual will relapse into crime, thereby assessing recidivism risk using socio-demographic, psychological, and behavioral features. A synthetic dataset was developed to simulate real-world patterns while maintaining appropriate ethical considerations: examples of covariates include age, gender, count of prior offenders, substance use history, social support, and psychological assessments. Several ML models were developed, trained, and compared with each other including Logistic Regression, XGBoost, Support Vector Machines (SVM), K-Nearest Neighbour’s (KNN), and Random Forest. With the goal of identifying individuals at higher risk of reoffending, the SVM model showed the best predictive performance. To reduce bias, Fairness-aware techniques were used addressing shortcomings in conventional systems, this scalable approach provides a framework that offers fair and objective standard for parole decisions, rehabilitation planning, and policy development. Future research will expand the use of explainable AI techniques to improve transparency and validate findings with criminal justice data in the real world.

Recidivism, Predictive Modeling, Risk Assessment, Criminal Justice, Fairness, Rehabilitation, Parole, Bias Mitigation

"PREVENTING RECIDIVISM USING MACHINE LEARNING", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.12, Issue 5, page no.156-163, May-2025, Available :http://www.jetir.org/papers/JETIRGV06022.pdf

"PREVENTING RECIDIVISM USING MACHINE LEARNING", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.12, Issue 5, page no. pp156-163, May-2025, Available at : http://www.jetir.org/papers/JETIRGV06022.pdf