Latent models, called hidden Markov models (HMMs), are types of algorithms that have been designed to detect crime activities by obtaining a sequence of observations from hidden values. The main contribution of these types of models is the fusion of coupled parameters with two types of HMM algorithms. The first algorithm is the Viterbi algorithm, which is commonly used to find the most probable path, and the accuracy of this algorithm is equal to 80%. The second algorithm is the Baum?Welch algorithm, which has been used to produce robust and accurate models. The modeling results normally focus on evaluating relative mean square errors in log likelihoods, transition matrices, and emission matrices for comparison of modeling performance based on different tolerance values. Previous reports have shown that the modified Baum?Welch algorithm can achieve good results for decreasing tolerance values. The goal of this Work is to generate a compact model that deals with ternary parameters rather than binary parameters by determining the sequential relation of past crime types and locations. Geographic locations can improve the HMM visualization in MATLAB. Moreover, crime levels and their most probable locations are predicted. The obtained results prove the goal of this work.
展开▼
