Agent-based modelling with geographically weighted calibration for intra-urban activities simulation using taxi GPS trajectories

Human motivations are an important factor in influencing human movement. However, most existing studies on passenger travel demand prediction focus on external characteristics of movement, but neglect the influence of activities and the motivations behind them, on the citizen's trip decisions. In this study, we proposed an agent-based model, to predict passengers' travel behaviour over a period of time, particularly when the urban structure changes. The model includes passenger characteristics, transitions in travel demands between activities over time, and their movement in space and time. In addition, we innovatively calibrated the agent-based model locally using Geographically Weighted Regression (GWR) to account for geographical variations in the parameters of destination attractiveness and travel cost in the agent-based model. We conducted a case study in Ningbo, China, using trip diaries, census data, and over 1.5 million taxi trip records. Our agent-based model showed superior performance in predicting citizens' movements and activities after a new shopping area in Ningbo was built, compared with a model without local calibration. The results also revealed the geographic sensitivity to destinations and the effects of a passenger's motivations that underpin human movement.