A multi-task deep learning model for short-term taxi demand forecasting considering spatiotemporal dependences

Abstract

Short-term taxi demand forecasting is of great importance to incentivize vacant cars moving from over-supply regions to over-demand regions, which can minimize the wait time for passengers and drivers. With the consideration of spatiotemporal dependences, this study proposes a multi-task deep learning (MTDL) model to predict short-term taxi demand in multi-zone level. The nonlinear Granger causality test is applied to explore the causality relationships among various traffic zones, and long short-term memory (LSTM) is used as the core neural unit to construct the framework of the multi-task deep learning model. In addition, several hyperparameter optimization methods (e.g., grid search, random search, Bayesian optimization, hyperopt) are used to tune the model. Using the taxi trip data in New York City for validation, the multi-task deep learning model considering spatiotemporal dependences (MTDL∗) is compared with the single-task deep learning model (STDL), the full-connected multi-task deep learning model (MTDL#) and other benchmark algorithms (such as LSTM, support vector machine (SVM) and k-nearest neighbors (k-NN)). The experiment results show that the proposed MTDL model is promising to predict short-term taxi demand in multi-zone level, the nonlinear Granger causality analysis is able to capture the spatiotemporal correlations among various traffic zones, and the Bayesian optimization is superior to the other three methods, which verified the feasibility and adaptability of the proposed method.