The rapid growth of transportation network companies (TNCs) has reshaped the traditional taxi market in many modern cities around the world. This study aims to explore the spatiotemporal variations of built environment on traditional taxis (TTs) and TNC. Considering the heterogeneity of ridership distribution in spatial and temporal aspects, we implemented a geographically and temporally weighted regression (GTWR) model, which was improved by parallel computing technology, to efficiently evaluate the effects of local influencing factors on the monthly ridership distribution for both modes at each taxi zone. A case study was implemented in New York City (NYC) using 659 million pick-up points recorded by TT and TNC from 2015 to 2017. Fourteen influencing factors from four groups, including weather, land use, socioeconomic and transportation, are selected as independent variables. The modeling results show that the improved parallel-based GTWR model can achieve better fitting results than the ordinary least squares (OLS) model, and it is more efficient for big datasets. The coefficients of the influencing variables further indicate that TNC has become more convenient for passengers in snowy weather, while TT is more concentrated at the locations close to public transportation. Moreover, the socioeconomic properties are the most important factors that caused the difference of spatiotemporal patterns. For example, passengers with higher education/income are more inclined to select TT in the western of NYC, while vehicle ownership promotes the utility of TNC in the middle of NYC. These findings can provide scientific insights and a basis for transportation departments and companies to make rational and effective use of existing resources.

中文翻译：

---

建筑环境对纽约市出租车和乘车游乐设施的时空变化影响

运输网络公司（TNC）的快速发展已经改变了全球许多现代城市的传统出租车市场。这项研究旨在探讨传统出租车（TN）和TNC上建筑环境的时空变化。考虑到骑行分布在空间和时间方面的异质性，我们实施了地理和时间加权回归（GTWR）模型，并通过并行计算技术对其进行了改进，以有效评估两种模式下本地影响因素对每月骑行分布的影响在每个出租车区。在纽约市（NYC）进行了案例研究，使用了TT和TNC从2015年至2017年记录的6.59亿个接载点。来自四个组的14个影响因素，包括天气，土地使用，社会经济和交通，被选为自变量。建模结果表明，改进的基于并行的GTWR模型比普通最小二乘（OLS）模型具有更好的拟合结果，并且对于大型数据集更有效。影响变量的系数进一步表明，在下雪天气中，TNC更加方便了乘客，而TT更集中在靠近公共交通的位置。此外，社会经济属性是造成时空格局差异的最重要因素。例如，受过高等教育/收入较高的乘客更倾向于在纽约州西部选择TT，而拥有汽车则促进了TNC在纽约市中部的效用。