Abstract:
With the arrival of Big Data Age, how to mining useful information hidden behind big data has become a hot topic in Geographic Information Science (GIScience). This study examined the spatio-temporal variation of regional taxi flows by analyzing a Big Data set of taxi trips in New York City (NYC) and the research combined the knowledge and methods of spatial statistics and visualization to extract the variation in both spatial and time dimensions. First, the spatial pattern of the overall taxi volume was analyzed by the Moran’s I Local Indicators of Spatial Association (LISA) to measure where the clusters are and how they vary from place to place each one-hour time segment of one day. In addition, the spatio-temporal pattern of the netflow was visualized in 3D-GIS environment to measure where “source” areas are, as well as using Empirical Orthogonal Function (EOF) to extract the primary information of the outgoing flow (outflow). Lastly, in order to explore the year level pattern, multivariate mapping was established to compare the areas with high/low outflows in 2015 with those in 2012 and 2009, respectively. The final results revealed a more specific understanding of regional taxi flows including the patterns of the overall flow, netflow, and outflow in NYC.
