Enhanced Crime Prediction: Leveraging CNN-LSTM Fusion for Improved Accuracy and Temporal Pattern Recognition

Abstract

Urbanization is a contributing factor to numerous social problems. Crime is one of these issues that exists in every city on the planet. A significant amount of data is gathered by police databases, which may be examined to lower the rate of crime. The investigation of criminal activity and the estimation of several crimes ruins one of the most exciting complications for investigators. It is not unusual for people in developing nations like India to hear about crimes occurring frequently. With the quick development of cities, we have to be continually conscious of our surroundings. To effectively manage such expanded and intelligent crimes, it is significant to investigate the recent crime trends and make organized and inclusive countermeasures to encounter the new movements of crime. Preventing crimes before they occur is always preferable to dealing with them after they have already occurred. Because of the recent dramatic advancements in machine learning technology, research on data analysis and crime prediction systems is crucial to lowering the rate of crime. The hybrid prediction method observes crime rates in order to avoid the unfortunate. A hybrid model based on deep learning approaches that combines a convolutional neural network (CNN) model and a long-short-term memory (LSTM) model to improve the accuracy of crime rate prediction is proposed. The CNN layers are added first in this CNN-LSTM method, and the LSTM layers are then added. On the procedure of the CNN model for feature extraction, and subsequent to the LSTM method to understand the features through time steps that have a high density layer for output,. The CNN and LSTM models provide a complete guide for crime rate analysis. The performance of the CNN and LSTM models is provided with 97.8% accuracy. When compared to the traditional methods, the proposed method yields high accuracy.