CRIME FORECASTING WITH MACHINE LEARNING AND DEEP LEARNING: A SYSTEMATIC ANALYSIS

This study analyzes and predicts crime patterns using a dataset of crimes from Chicago starting from 2001. Through exploratory data analysis (EDA), we uncovered unique and non-unique case numbers and crime distributions. The data was preprocessed for machine learning by handling duplications, extracting date components, and encoding categorical data. Visualization techniques highlighted key trends and high-crime areas. We developed a predictive model using an artificial neural network (ANN), equipped with dense and dropout layers to prevent overfitting, and optimized using the Adam optimizer for regression tasks. The model's effectiveness was assessed with mean squared error (MSE) and mean absolute error (MAE), confirming its ability to accurately predict crime counts. Additionally, we integrated the model into a Gradio web interface, allowing easy user interaction for crime count predictions. This research enhances understanding of crime dynamics and offers a practical tool for resource allocation in public safety, demonstrating the impact of machine learning and deep learning in predictive analytics within the public safety sector.