Why Deep Learning Is Not Necessary for Traffic Forecasting?

Deep learning is not simply an overal better tool. And therefore, it should not be applied blindly for any task. If you don't obtain incitful conclusions for an improvement, why are you using it? Good question. Is as certainly a good hypothesis. We have seen things like translation models, for example, go leaps and bounds better when we applyd deep learning. Perhaps they could for traffic systems as well.