Artificial Intelligence Aspect Of Transportation Analysis Using Large Scale Systems

Problem: The problem of the finalized exploration revolved around the inadequacy of current traffic forecasting models. Despite decades of examination in many fields, existing approaches, often relying on linear models and stationary time series assumptions, struggle to accurately predict traffic under chaotic events. The identified limitation has profound consequences, evident in the significant economic losses and time inefficiencies incurred due to traffic congestion, as exemplified by the $144 billion in losses and the 34% increase in travel time for drivers in Los Angeles County in 2013. The challenge lies in the inherently unpredictable nature of traffic events, ranging from regular rush hours causing sharp declines in traffic speed to unpredictable accidents leading to unforeseen delays. Consequently, there is a pressing need for a more effective and adaptive traffic forecasting model that can reliably operate under both normal and abnormal traffic conditions, addressing the shortcomings of traditional linear models and stationary time series assumptions. Purpose: The purpose of the completed investigation was to determine whether a traffic forecasting model that incorporates machine learning and deep learning technologies can yield effective traffic forecasts based on real-time weather and traffic data. Method: The study involved the development of a traffic forecasting model as informed by existing literature. Data collection was done through simulating data similar to the traffic data from Los Angeles County that was utilized in Yu et al.'s research on deep learning and traffic prediction in extreme weather scenarios [1]. Data analysis was done through MAE and t-test. Results: The findings demonstrated that the created traffic forecasting model outperformed the current methods in its ability to provide traffic forecasts with a better degree of accuracy. Conclusion: Regardless of the traffic volume, weather, or time of day, the developed traffic forecasting algorithm can give precise real-time traffic predictions.