Short-Term Traffic Flow Prediction: Using LSTM
Pregya Poonia, V. K. Jain
Abstract
Traffic data is being exploded in past few years and that is because of the increasing number of vehicles. People get struck in the traffic for hours so, accurate flow of traffic is really important for both the traveler and intelligent transportation system. Existing models somehow fails to provide accurate information of flow and that is because they are using shallow forecast models which are as yet unsatisfying for real-time applications. This circumstance makes us to consider the issue dependent on profound design models. In this paper, we have applied the utilization of Long Short-Term Memory Networks (LSTM) for momentary traffic stream forecast. LSTM is a deep learning approach which is capable of learning long-term dependencies and non-liner traffic flow data. It remembers the information for a long period of time which settles on it an appropriate decision in rush hour gridlock estimating. We have tested this model on continuous traffic informational collections and got great execution of our model.