Modelling ride-sourcing matching and pickup processes based on additive Gaussian Process Models

Matching and pickup processes are core features of ride-sourcing services. Previous studies have adopted abundant analytical models to depict the two obtain operational insights; while goodness fit between data was dismissed. To simultaneously consider fitness analytically tractable formations, we propose a data-driven approach based on additive Gaussian Process Model (AGPM) for market modeling. The framework is tested real-world collected in Hangzhou, China. We models, machine learning AGPMs, which number matches or pickups used as outputs spatial, temporal, demand, supply covariates utilized inputs. results demonstrate advantages AGPMs recovering terms estimation accuracy. Furthermore, illustrate modeling power AGPM by utilizing trained model design estimate idle vehicle relocation strategies.