Space–Time Analysis: Concepts, Quantitative Methods, and Future Directions

Abstract

Throughout most of human history, events and phenomena of interest have been characterized using space and time as their major characteristic dimensions, in either absolute or relative conceptualizations. Space–time analysis seeks to understand when and where (and sometimes why) things occur. In the context of several of the most recent and substantial advances in individual movement data analysis (time geography in particular) and spatial panel data analysis, we focus on quantitative space–time analytics. Based on more than 700 articles (from 1949 to 2013) we obtained through a key word search on the Web of Knowledge and through the authors' personal archives, this article provides a synthetic overview about the quantitative methodology for space–time analysis. Particularly, we highlight space–time pattern revelation (e.g., various clustering metrics, path comparison indexes, space–time tests), space–time statistical models (e.g., survival analysis, latent trajectory models), and simulation methods (e.g., cellular automaton, agent-based models) as well as their empirical applications in multiple disciplines. This article systematically presents the strengths and weaknesses of a set of prevalent methods used for space–time analysis and points to the major challenges, new opportunities, and future directions of space–time analysis.

大部份人类历史所关注的事件与现象，透过运用空间和时间作为其主要的特徵面向，以绝对或相对的概念化进行描绘。时空分析企图理解何时、何地（有时是为何）事情会发生。在个体活动数据分析（特别是时间地理学）和空间面板数据分析的部分最晚近且实值进展的脉络中，我们聚焦量化时空分析。根据我们对知识网进行关键词搜索以及作者的个人档案资料所获得的七百篇以上的文章（自1949 年至 2013 年），本文对时空分析的量化方法，提供了综合性的概要。我们特别凸显时空模式揭露（例如各种集群计量、路径比较指标、时空检定）、时空统计模型（例如存活分析、潜在轨迹模型），以及模拟方法（例如细胞自动机、以行动者为基础的模型），以及它们在多重领域的经验应用。本文系统性地呈现一组用来进行时空分析的盛行方法的优劣之处，并指出时空分析的主要的挑战、崭新的契机，以及未来的趋势。

Durante la mayor parte de la historia humana, la caracterización de eventos y fenómenos interesantes se ha basado en el espacio y el tiempo como sus principales dimensiones, tanto en absolutas como relativas conceptualizaciones. El análisis espacio-tiempo busca comprender cuándo y dónde (y algunas veces por qué) ocurren las cosas. Dentro del contexto de varios de los más recientes y sustanciales avances en el análisis de datos del movimiento individual (geografía del tiempo en particular) y análisis de datos del panel espacial, nosotros nos enfocamos en la analítica cuantitativa del espacio-tiempo. Este artículo entrega una visión de conjunto sintética acerca de la metodología cuantitativa para el análisis del espacio-tiempo, a partir de más de 700 artículos (de 1949 a 2013) que obtuvimos por medio de una búsqueda con palabras clave en la Web of Knowledge [Web del Conocimiento] y en los archivos personales de los autores. En particular, destacamos el patrón de revelación del espacio-tiempo (e.g., varias medidas de agrupamiento, índices de comparación de rutas, pruebas de espacio-tiempo), modelos estadísticos de espacio-tiempo (e.g., análisis de supervivencia, modelos de trayectoria latente), y métodos de simulación (e.g., autómata celular, modelos basados en agente) lo mismo que sus aplicaciones empíricas en múltiples disciplinas. Este artículo presenta sistemáticamente las fortalezas y debilidades de un conjunto de métodos prevalentes usados para el análisis del espacio-tiempo y apunta a los principales retos, nuevas oportunidades y direcciones futuras del análisis espacio-tiempo.

Key Words:

• absolute versus relative space
• review
• simulation models
• space–time analysis
• statistical models

关键词：:

• 绝对空间相对于相对空间，回顾，模拟模型，时空分析，统计模型。

Palabras clave:

• espacio absoluto vs. espacio relativo, modelos de simulación, análisis espacio-tiempo, modelos estadísticos

Acknowledgments

We are grateful for the very helpful comments from the anonymous reviewers and the editor, Dr. Mei-Po Kwan. We thank San Diego State University for ongoing support of our research. Thanks also go to Elias Issa and Evan Casey for their assistance in data compiling and analysis.

Funding

We are indebted for the financial support from the National Science Foundation (Award #1028177 and DEB-1212183).

Notes

1. We consider geographic space only in the context of this article. Other dimensions of space, such as information space in relation to channel capacity, entropy, and information gain (Shannon and Weaver 1964), are not considered.

2. The term distance, unless otherwise specified, refers to the Euclidean distance in this article, among many other alternatives such as the least cost path, Manhattan, and network distances.

3. Space–time cube can be also used in representing, mapping, and understanding the arts and humanities (e.g., Travis 2014).

4. For this set of key words, if “title” is chosen, 618 records are returned, representing 36 percent of the 1,723 records returned by choosing “topic.”

5. As mentioned later in the Conclusion and Discussion, our search is based on Web of Knowledge alone and does not include all publications in the relevant disciplines. For instance, our search has found sixty-seven papers in social sciences; under the same search parameters, 102 papers published in scholarly peer-reviewed social science journals were returned, using the search engines of Communication and Mass Communication Complete (CMCC) and PsycINFO (Psychology).

6. Here the word continuously is not used in the strict mathematic sense but implies that people's movement can be potentially tracked in very fine spatial and temporal resolutions.

7. ABMs are also powerful in analyzing other nonhuman individual movement data. See Tang and Bennett (2010) for a nice review of agent-based modeling of animal movement.

8. In addition to the dichotomy of absolute versus relative space (or absolutism vs. relationalism), useful dimension for space–time data is the dichotomy of realism versus idealism, which refers to whether space–time or objects are mind-independent (e.g., physical objects) or mind-dependent (e.g., abstract constructs). See more detail in Yuan, Nara, and Bothwell (2014).