Accurately mining the spatiotemporal distribution characteristics of demand for taxi travel is helpful to better dispatch and guide the distribution of taxi supply, so as to alleviate the imbalance of taxi supply and demand in high passenger demand areas. Based on the multi-source traffic data including taxi GPS data, taximeter data, public transport transactions data and Point of Interesting (POI ) data in Beijing, correlation analysis methods were used to select the influencing factors of taxi travel demand, and establish a multi-dimensional set of influencing factors. A Geographical Weighted Regression (GWR) based influence model of regional taxi travel demand is established, and 1398 regions in Beijing is taken as an example to quantitatively explore the impact of various factors on taxi demand under different space-time conditions. The results show that the density of high-grade residential area, financial and commercial land, office area and recreational area in the city periphery has an obvious positive impact on taxi travel demand; While the density of residential land in urban periphery and the office area in the city center has a negative correlation with the taxi travel demand; In addition, the impact of the public transport volume in the peak hours on the taxi travel demand in the city center and peripheral areas is significantly different.The analysis of the distribution characteristics of taxi travel demand in different space-time dimensions provides an important support for the rational allocation of taxi transportation service resources.

Eco-routing has been proposed as a means of distributing traffic in cities to improve mobility sustainability [1-3]. The implementation of eco-routing in real-life requires a diverse set of information, including heterogeneous and legacy sensors often already present in the city infrastructure. In this work, we present a modular architecture leveraging Internet of Things (IoT) technologies that enables collecting the necessary data, fusing it, and inferring the information required for the eco-routing application. Further, we formulate the eco-routing problem as a multi-objective optimisation to distribute traffic targeting better pollutant emissions vs travel time trade-offs. A city manager chooses the desired solution, which is used to serve routes, e.g. to a fleet committed or incentivized to contribute to an environmentally friendlier city. Preliminary results show the potential impact of eco-routing using real data for a mid-sized European city, and the impact of using static emission weights in the optimization formulation.