Crime Rate Analysis and Mapping from Socio-Economic Data Using Deep Neural Networks

Crime prediction is the attempt to identify and mitigate future criminal activity. Crime is typically "unpredictable"; it cannot be predicted in advance. Addressing the roots of crime has long been a priority for researchers. Extracting causality from data is difficult if the relevant data aspects are not selected. In this research activity, an innovative classification framework is defined which is the enhanced future crime prediction. The framework consists of 3 major steps, the first step is the pre-processing step which is carried out based on the Lagrange polynomial interpolation method for filling the missing values. Next, the second step is the feature selection process which is employed using the Whale optimization algorithm. Feature selection is often regarded as a crucial step in any pattern identification procedure. Its purpose is to reduce the amount of memory used, the processing time, and the computational overhead of the classification process in order to improve the classification efficiency. The last step is the Classification process, in which the suggested feature selection and Whale optimization algorithm method will be used for selecting the important socioeconomic data based on their influence on crime rate prediction using the deep learning algorithm. Finally, the crime rate is analyzed and categorized as less prone, medium prone, and high prone with regard to crime activities. Several measures are used to analyze the suggested methodology's performance. This newly developed model is compared with existing models like Genetic algorithm, FireFly, and particle swarm optimization in terms of diverse performance metrics like execution time, memory consumed, training and testing time, sensitivity, specificity and accuracy. From the results, this model is proposed as the best crime prediction model compared with the other existing models. In comparison to other methods now in use, this approach effectively retrieves the crime's attributes with a high accuracy of 93.39%.