Have you ever been stuck in a sudden traffic jam for over an hour? If you knew that there would be heavy traffic, you would’ve taken an alternate route to save time.
You can build a traffic jam prediction model using deep learning techniques in Python. You can use openly available Waze datasets for this purpose. You can get data of various traffic event types, along with their date, time, and exact location. You can then build a model that predicts the location and time of the next traffic jam.
There are many existing models and research papers on this topic that you can read, and many different methodologies have managed to produce high results.
One approach that managed to win a hackathon was the use of RNNs to predict severe traffic jams. Waze data was used to identify heavy traffic events. Then, a sequence of events leading up to the traffic jam was recorded along with their timestamps to train an RNN.
The model was built purely in Python with the Keras library and delivered highly accurate results.
Dataset: Waze Open Dataset
