In modern society, the demands of sustainable urban development and the diverse needs of the populace are increasing. One of the approaches gaining attention to realize such urban planning is simulations based on urban data. However, traditional approaches often treat inhabitants as homogeneous entities. While some research has attempted to take in the complexity of individual behaviors, traditional studies have primarily modeled consumer utility using linear polynomials, which presents mathematical limitations. This study introduces a transdisciplinary approach that combines social scientific efforts to replicate complex urban conditions with engineering methods, particularly time series prediction, for simulating the decision-making processes of agents within a simulation. Utilizing urban pedestrians’ data, the model constructs predictions for subsequent actions, adapting these predictions to enhance the decision-making models of agents in the simulation. The validation of this model, assessing its ability to replicate actual decision-making behaviors of individual users, indicates a promising level of reproducibility. This study provides significant insights for governmental agencies and urban developers, contributing to more efficient and effective urban planning and development strategies. Achieving sustainable urban development in this manner ensures the well-being of urban populations and the long-term viability of urban environments, demonstrating the model’s potential to inform and enhance urban planning efforts.