| This dissertation presents an exploratory spatial data analysis (ESDA) for origin and destination flow data (O-D flow data). In a number of problem domains, there is an increasing interest in exploring flow data. This interest is largely as a result of the advancement of computer technology both in software and hardware and generated by the need and possibility to analyze the large amount of spatial data available. Flow data, which is defined as data that captures movement between origins and destinations, may consist of movement of message traffic over internet, freight flows, commodities moving from state to state, population migrations, international trade, patient referrals to hospital, shopping patterns, and diffusion of disease or of innovations. Looking for patterns and anomalies of these flows is an important factor in the statistical modeling of the social and economic structure of a system of regions. However, exploration and analysis of these flows (i.e., describing and visualizing their spatial distribution and association, discovering patterns of clustering, and suggesting different spatial regimes) has been hindered by a lack of appropriate tools.; The focus of this dissertation lies in the research tradition of ESDA, which requires explicitly the existence of geographical attributes in the datasets, interactive visualization tools and dynamic graphics. In this study, I propose a new set of techniques for exploring O-D flow data and integrate it into an operational ESDA environment. These objectives are decomposed into the following components. First, an ESDA framework for O-D flow data is proposed by extending it from traditional ESDA frameworks. There are two important components in the framework: (1) Interactive Visualization; and (2) Data Structures. The former consists of a set of interactive visualization techniques and dynamic graphics (devices) for conducting exploration of O-D flow data. The later provides basic infrastructures for coordinating data and devices, and communicating between user-interface, data and devices (i.e., linking and brushing). Third, an easy-to-use prototype software system is implemented for the exploration and visualization of flows. Finally, the use of the 48 US state migration during 1995-200 demonstrated and tested the system and methods developed with real flow data in order to proof-of-concept result. |
