Traffic Accident Prediction System

The growing rate of urbanization and the surge in vehicular traffic have significantly increased the frequency of road accidents, leading to loss of life, property damage, and substantial economic consequences. In response to these challenges, this project proposes a Traffic Accident Prediction System, an intelligent, data-driven solution that utilizes machine learning to predict the likelihood of traffic accidents based on a range of environmental and contextual factors. The system employs supervised learning techniques, training models on a real-world dataset comprising labelled records of past accidents. It incorporates critical features such as weather conditions, road surface types, lighting conditions, vehicle categories, and time of day, traffic density, speed limit, number of vehicles, driver alcohol, accident severity, road condition, driver age, driver experience. After comprehensive data preprocessing—including cleaning and handling of missing values— several machine learning algorithms, including Random Forest, Decision Tree, Logistic Regression, Support Vector Machine, Naive Bayes, K Nearest Neighbor are evaluated. The model with the highest performance, measured through accuracy, precision, recall, and F1-score, is selected for deployment. The system delivers three primary functionalities: (1) Accident Risk Prediction, which estimates the probability of an accident based on real-time input data; (2) Real-Time Input Handling, allowing users to submit current traffic and environmental conditions through a web interface; and (3) Decision Support Output, which provides interpretable predictions to assist traffic authorities, planners, and drivers in making proactive safety decisions.