Crime Prevention and Detection of Using Predictive Algorithms: A Case Study of Mangu, Riyom and Bokkos Metropolis in Plateau State

For law enforcement organizations around the world, preventing and detecting crimes are crucial problems. Predictive algorithms are a possible method of spotting crime trends and improving public safety in light of the growing amount of crime data available and the developments in machine learning. The city of Plateau and around it environs has seen an increase in criminal and herdsmen activity, but intelligent innovative solutions to guide security forces in combating the threat are lacking. The study seeks to achieve the modeling of crime prevention and detection using predictive algorithms method. Specific objectives to be achieved are to model crime detection and prevention, analyse prediction algorithms, compare the performance of these algorithms and evaluate the performance of the proposed model. The general methodology employed was the quantitative research approach. Data was collected from security agencies within Mangu, Bokko, Riyom L.G.A of Plateau state, with an additional dataset gotten online. Three supervised machine learning algorithms of Decision Trees, Random Forest, and Support Vector Machines (SVM) were used for the study. The analysis of the models was achieved using Python programming language, alongside some libraries of it. The models were evaluated using accuracy, precision, recall, and F1-score. For accuracy, 1.000, 0.9950 and 0.5075 was obtained; precision was 1.000, 0.9955, 0.5372; recall achieved 1.000, 0.9950, 0.5075, and F1-score values were 1.000, 0.9951 and 0.4910 for decision trees, random forest and support vector machines respectively. The findings showed that both Random Forest and Decision Trees performed exceptionally well, with Decision Trees achieving perfect scores on every metric (1.0000). SVM, on the other hand, performs poorly, emphasizing how crucial it is to use the right algorithms for crime prediction tasks. Law enforcement organizations can better allocate resources and put targeted preventive measures into place by using visualizations of crime frequency by location, time of occurrence, and victim age distribution. The study notes a number of research gaps in spite of these encouraging findings, such as the requirement for real-time crime prediction systems, the incorporation of outside data sources, sophisticated geospatial and temporal analysis, and ethical considerations in intelligent predictive policing.