ML engineering project

Crime Prediction

Crime prediction is a critical task for law enforcement agencies and urban planners to allocate resources effectively and prevent criminal activities. Chicago, being one of the largest cities in the United States, faces various challenges related to crime, making it an ideal candidate for such predictive modeling efforts. **Research Foundation:** This project implementation is based on the research paper ["Crime Prediction Using Machine Learning"](https://arxiv.org/pdf/2303.16310.pdf), which explores various ML techniques for anticipating criminal activity patterns in urban environments.  **Data Sources:** The data used in this project consists of historical crime records obtained from the [Chicago Police Department](https://data.cityofchicago.org/Public-Safety/Crimes-2001-to-Present/ijzp-q8t2/data), containing over 20 years of crime incidents from 2001 to present.   **Methodology:** • **Data Collection** - Gather historical crime data from the Chicago Police Department and relevant socio-economic and environmental datasets • **Data Preprocessing** - Clean the data, handle missing values, and engineer features • **Feature Selection** - Identify significant features that contribute to crime prediction, including time of day, day of week, and location characteristics • **Model Development** - Train various machine learning models including Support Vector Machine and Random Forest Regression • **Model Evaluation** - Evaluate model performance using metrics like accuracy, precision, recall, and F1-score **Key Model Features:** The predictive models leverage temporal and spatial features: • **Temporal Features**: Time of day, day of week, seasonal patterns • **Spatial Features**: Geographic coordinates, neighborhood characteristics • **Historical Patterns**: Crime type frequencies, trend analysis **Prediction Results:**    **Deployment:** The best-performing model is deployed as a web application with Flask backend, allowing users to input various parameters and receive real-time crime predictions for different locations in Chicago. The application features interactive Bing Maps visualization for geographical crime analysis. **Tech Stack:** Built with **Python**, **Flask**, **scikit-learn**, **Pandas**, **NumPy**, **Jupyter**, and **Bing Maps API**. The system processes large-scale crime datasets and provides interactive visualizations for predictive analytics.