Hybrid Crime Detection System Using K-Means and Gradient Boosting
i-manager's Journal on Digital Forensics & Cyber Security · 20.12.2026 · 4 (2) · 1
DOI: 10.26634/jdf.4.2.1002 ↗Kurz gefasst
Fachartikel in der Zeitschrift i-manager's Journal on Digital Forensics & Cyber Security, online erschienen am 15.12.2026. Verfasst von Emmanuella Lubinda und 2 weiteren Autor:innen.
Abstract (Originalsprache)
Crime remains one of the most significant challenges affecting public safety and socioeconomic development worldwide. Traditional crime analysis approaches primarily rely on historical investigations and manual interpretation of crime records, making it difficult to identify hidden crime patterns and accurately predict future criminal activities. To address these limitations, this study proposes a Hybrid Crime Detection System using K-Means Clustering and Gradient Boosting (KM-GB) for crime hotspot identification and crime prediction. The proposed framework consists of two stages. First, the K-Means clustering algorithm is employed to analyze crime-related attributes, including location, time, and crime type, to identify crime hotspots and uncover hidden patterns within crime datasets. The resulting cluster labels are then incorporated into a Gradient Boosting model, which learns complex relationships among crime features and predicts the likelihood and location of future crime incidents. The integration of clustering and classification enables both descriptive and predictive crime analytics within a unified framework. Experimental evaluation was conducted using publicly available crime datasets. The performance of the proposed KM-GB model was compared with conventional machine learning techniques, including Decision Tree, Logistic Regression, Random Forest, and Support Vector Machine (SVM). The results demonstrate that the proposed hybrid model achieved an accuracy of 94%, outperforming the benchmark models in terms of precision, recall, and F1-score. Furthermore, crime hotspot visualization using geographical mapping provided interpretable insights for law enforcement agencies. The findings indicate that integrating unsupervised clustering with supervised learning significantly improves crime prediction accuracy and hotspot detection. The proposed KM-GB framework offers an effective decision-support tool for proactive crime prevention, resource allocation, and public safety enhancement.
Zitieren
APA
Lubinda, E., Mwenya, P., & Esther, J. (2026). Hybrid Crime Detection System Using K-Means and Gradient Boosting. i-manager's Journal on Digital Forensics & Cyber Security, 4(2), 1. https://doi.org/10.26634/jdf.4.2.1002
Vancouver
Lubinda E, Mwenya P, Esther J. Hybrid Crime Detection System Using K-Means and Gradient Boosting. i-manager's Journal on Digital Forensics & Cyber Security. 2026;4(2):1. doi:10.26634/jdf.4.2.1002
BibTeX
@article{Lubinda2026,
author = {Lubinda, Emmanuella and Mwenya, Patrick and Esther, J.},
title = {Hybrid Crime Detection System Using K-Means and Gradient Boosting},
journal = {i-manager's Journal on Digital Forensics & Cyber Security},
year = {2026},
volume = {4},
number = {2},
pages = {1},
doi = {10.26634/jdf.4.2.1002}
}
EUVIMED