Design and development of mobile crime risk assessment systems (ICRAS) for investigation officers using ETL based data processing and predictive analysis

Crime is a significant threat to public safety, social stability, and national development, particularly in rapidly urbanizing countries like Malaysia. Despite the Royal Malaysian Police (RMP) investing in crime reduction initiatives, the lack of a unified analytical framework hinders proactive crime prevention and resource optimization. Existing crime management strategies often rely on fragmented datasets, delayed reporting, and reactive decision-making, which limit early intervention efforts. This study aims to address these challenges by developing the Integrated Crime Risk Assessment System (ICRAS), a data-driven platform designed to enhance crime risk evaluation and management for law enforcement agencies. The primary objective is to consolidate multiple data streams, crime records, socio-economic indicators, spatial patterns, and environmental variables into a unified system for more effective crime risk assessment. The research employs an Extract, Transform, and Load (ETL) pipeline to clean, standardize, and merge the data, ensuring analytical accuracy and regional comparability. The system integrates two core analytical tools: the Risk Appetite Index and the Crime Risk Matrix, which quantify the likelihood and severity of crime incidents. To enable predictive capabilities, ICRAS utilizes the Random Forest algorithm, chosen for its robustness, interpretability, and ability to handle complex, nonlinear data. This model identifies key predictors of criminal activity and provides short-term forecasts to inform proactive policing. The dataset for this study includes 5,315 cases of violent and property crime in Malaysia from 2015 to 2021. Pilot testing conducted in Selangor showed that the Risk Appetite Index achieved a readability score of 4.52 and a Cronbach’s α reliability coefficient of 0.86, while the Crime Risk Matrix scored 4.47 and 0.83, respectively. Both tools surpassed established thresholds for internal consistency and user comprehension. Furthermore, the implementation of ICRAS demonstrated improved investigative accuracy, reduced redundancy, and shortened decision-making cycles. Officers reported that the system’s color-coded risk visualizations and structured descriptors enhanced interpretation, while its predictive insights supported faster, more confident judgments. ICRAS presents a methodologically sound and operationally viable solution for modern policing in Malaysia. By integrating ETL-based data processing, Random Forest prediction, and interactive visualization, this framework provides law enforcement agencies with a scalable, evidence-based tool for crime risk assessment. The findings not only offer a national solution but also present a transferable model for other jurisdictions seeking to institutionalize data-driven decision-making and enhance the analytical capabilities of their criminal justice systems.