Analysis of deep learning based algorithms for traffic parameters prediction using spatio temporal data

Abstract

There has been a rapid increase in traffic demand, due to the newlinedevelopment of urban areas and dense population. An efficient and intelligent newlinetraffic management system is essential to handle the growing volume of newlinevehicles and the expanding urban population within smart cities. Enhancing the newlinetraffic management system necessitates an accurate prediction of future traffic newlineconditions. The traffic data are spatiotemporal, highly complex, and nonlinear newlinein nature. The traffic flow, speed, and density are considered as the major traffic newlinecharacteristics. Traffic forecasting is a form of time series forecasting which newlineinvolves the forecasting of numerous traffic parameters such as flow, speed, newlineand density. This forecasting helps the traffic management system to reduce newlinecongestion. newlineThe traffic data are both geographically and temporally correlated. newlineHence, there is a need to consider both spatial and temporal features while newlinepredicting the traffic parameters. The traffic flow or congestion at a particular newlinelocation not only depends on a single parameter but also it depends on multiple newlinevariables such as traffic flow, speed, occupancy as well as other external factors newlinesuch as meteorological data, incident data. Hence, predicting the traffic data newlinebased on single parameter is not enough to provide an accurate prediction. newline