This research project delves into a pivotal aspect of intelligent transportation systems: the prediction of road accident severity for road safety. Employing machine learning methodologies such as support vector machines (SVMs), decision trees, multilayer perceptrons (MLP), stacked autoencoders, and LSTM models, we aimed to provide accurate predictions in both domains by integrating spatiotemporal data. The dataset encompasses comprehensive spatiotemporal information, including location, time, road characteristics, and weather conditions. Through training and evaluating the models on this dataset, we assessed their performance in accurately predicting accident severity and traffic volume. The findings indicate that MLP outperforms SVM and decision tree models in predicting accident severity, showcasing its adeptness in capturing intricate data relationships. Conversely, the stacked autoencoder model demonstrates superior performance in traffic volume prediction, leveraging its ability to extract deep features and capture temporal dependencies. This research contributes to the field of intelligent transportation systems by furnishing precise predictions for accident severity and traffic volume using advanced machine learning techniques. These predictions hold the potential to enhance road safety measures, optimize traffic management strategies, and facilitate informed decision-making in transportation infrastructure planning and optimization endeavors.