EGFD-AROMA-based Pattern Forecasting with Crime Rate Level Identification

In numerous cities, crime has become a chief concern that leads to the rising demand for timely predicting citywide crime occurrence. Therefore, this paper proposes an efficient forecast pattern with crime rate level identification. Primarily, the input crime video undergoes preprocessing. Afterward, the features are extracted. Then, by utilizing Permuted Congruential Generator-based Northern Goshawk Optimization (PCG-NGO), the important features are selected. Thereafter, feature similarities are identified. Conversely, the attributes are extracted from the crime detail dataset. Next, by considering the selected features, identified similarities, and extracted attributes, the crime categories are clustered using the JaroWrinkler distance-based K-Means Algorithm (JW-KMA). Then, the dependence strength is analyzed for the clustered categories. Lastly, for predicting patterns and crime rate levels, a Fuzzy Inference System and Convolutional Encoder Decoder-based Residual Neural Network (FIS-CEDRNN) is employed. By utilizing the extracted attributes from the crime detail dataset and the predicted patterns, Entropy Generalized Finite Difference-based Autoregressive Integrated Moving Average (EGFD-ARIMA) is trained for forecasting crime patterns. The experimental outcome exhibited that the proposed technique withstands maximum accuracy when contrasted with existing methodologies.