Optimizing taxi parking spaces in urban environments is crucial for reducing customer wait times and enhancing operational efficiency. Traditional approaches, based on static statistical methods and relying on Points of Interest (PoIs), often fail to address dynamic demand patterns, particularly with the rise of app-based ride-hailing services. This paper introduces a data-driven framework combining automatic segmentation and association rule mining techniques to dynamically align taxi parking spaces with demand zones. Using DBSCAN for parking cluster identification and the Apriori algorithm for correlating these clusters with high-demand areas, our methodology demonstrates a practical solution. Experimental results reveal positive alignment between identified parking spaces and manually validated demand areas, highlighting the reliability of our approach in modern urban transportation systems.

Accurate demand prediction in target areas is critical for an optimal taxi placement systems. This accuracy relies heavily on the correctness of the spatial index because predictions are directly runned on this spatial representation. However, conventional spatial index methods often fail to fully capture the dynamic patterns of urban demand, leading to inefficiencies in fleet management and service delivery. In this paper, we propose a novel methodology that integrates demand hotspots into an adjusted spatial grid. We demonstrate the implementation of this approach in a simulated environment and evaluate it against conventional spatial index. The results show that our proposed spatial index has more accurate representation of demand while having acceptable performance, leading to better taxi placement. This strategy is scalable and adaptable to various urban settings in the field of modern transportation.