Short-Term Forecasting of Emerging On-Demand Ride Services

In the last few years, on-demand ride services boomed worldwide, and different modes of ridesourcing services emerged, too. However, there have been few qualitative and quantitative analyses on these ride service patterns, partially due to the lack or unavailability of detailed on-demand ride service data. In this paper, we analyze the real-world individual-level order and the trip data extracted from the DiDi's on-demand mobility platform in Hangzhou, China. This study intends to understand the temporal and spatial travel pattern of passengers' demand and ride services which include four types, i.e., Taxi Hailing, Private Car Service, Hitch, and Express. We study the relationship between different service modes of the drivers from a selected region in specific time periods. In order to predict travel demand of the aforementioned on-demand ride services, we utilize LASSO (least absolute shrinkage and selection operator) to rank features of the on-demand platform data (e.g., distance, fee, and waiting time). An on-demand ride prediction model is established based on the random forest (RF), which is then compared with the autoregressive integrated moving average (ARIMA) and support vector regression (SVR). The results show that RF outperforms other models and it is utilized to provide an insight for forecasting the demand of distinctive on-demand ride service patterns. To the best knowledge of authors, this paper is among the first attempts to learn the temporal and spatial travel patterns, also to forecast emerging on-demand ride services.