Uber and other ride-hailing systems operate in spatiotemporal dynamics; having accurate ride demand prediction and fare pricing greatly influences customers’ experience as well as system efficiency. According to current research that splits Uber’s operations into demand forecasting and fare pricing, both tasks are affected by highly similar external factors that include time attributes, pickup and drop-off locations, distances between two points, weather, etc. This paper models both ride demand forecasting and fare price prediction with a Gradient Boosting Regressor (GBR)-based machine learning framework. Furthermore, we demonstrate that dynamic pricing can help the optimization of this objective so Uber can make better dispatching decisions of drivers and improve decision-making using intelligent ride-hailing considering demand and price together.
Rajeswari et al. (Thu,) studied this question.