Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
1 The objectives of this research were (a) to evaluate the extent to which cell phone data accurately reflect daily travel and (b) to develop guidelines on how to best use these data to understand and model travel behavior. Given the interest in new sources of locational data, it is critical to evaluate the strengths and weaknesses of this new stream of data and its applicability to the development and application of travel demand models. This report is for transportation practitioners and agency staff interested in the value of and potential applications for cell phoneâderived travel data. Other interested parties may include transportation modelers and planners and their peers in metropolitan planning organizations (MPOs), state departments of transportation (DOTs), and federal agencies. The research provides insights on the strengths and weaknesses of cell phoneâderived travel data and provides advice on how best to harvest its value. The Behavioral Paradigm and Disruptive Technologies New and innovative approaches to data and analytical methods may be indicative of disrup- tive technologies that provide opportunities and challenges to an industry. The emergence of locational data represents such a disruptive technology. The availability of new and detailed sources of vast amounts of travel data may signal a shift in how transportation planning and travel demand modeling will be conducted in the future. Although these data sources will affect how well society understands the present and how it should plan for the future, there are competing hypotheses about how these data will affect thinking and practices. One hypothesis is that these data streams are so powerful and detailed that they can replace the traditional approach to modeling, either in its entirety or for key components of the model system. Another hypothesis is that these new data streams will be harnessed to offer valuable inputs to understanding todayâs observed travel patterns, both for traditional and for new forms of transportation. The research team approached these new streams of data as unique opportunities to get better snapshots of travel patterns and to enhance its understanding of the travel behavior of todayâs drivers. The team believes that these new forms of data will help the evolution of models within a travel behavior framework to help create methodological advancements that will more accurately reflect todayâs travel patterns and will allow practitioners to evaluate what-if scenarios for more nuanced policy decision-making needs today and in the future. The question is how these new data sources will become part of planning and modeling practice over the next 10 years. Locational data in an aggregate form offer opportunities to support validation and allow for easier and more frequent model updates. Locational data S U M M A R Y Cell Phone Location Data for Travel Behavior Analysis
2 Cell Phone Location Data for Travel Behavior Analysis in a disaggregate form hold the promise of a better understanding of the drivers of demand for both traditional modes and the emerging sharing economy modes. These new locational data sources are expected to disrupt the way practitioners think about travel, and these sources can augment traditional data collection methods as long as sampling issues and biases are understood and addressed. The research team believes in the concept of creative collaboration, in which new locational data sources can augment and enhance thoughtful, behaviorally based approaches to planning and modeling. The research team believes that the value of new data sources will be enhanced if insights from locational data are incorporated into transportation practitionersâ understanding of the factors driving travel behavior. Approaches that integrate these new data streams into the behavioral paradigm for travel can fill gaps in existing models and allow for policy- sensitive and sophisticated approaches to answer todayâs nuanced policy questions. The individual traveler and the array of travel choices that this individual makes, coupled with the choices made by other members of the individualâs household, remain at the center of the behavioral paradigm describing daily travel. The key question is whether and how a better understanding of travel patterns for traditional and new modes can be gained from new sources of locational data. The current behavioral paradigm and modeling framework can benefit from new data to (a) become more policy-sensitive and (b) forecast traveler choices and travel patterns with greater confidence. The Types of Questions to Ask To meet the challenge of interpreting and applying these data, the research team used a case study approach that relied on data and models from Boston. The research team com- pared and contrasted traditional survey data and models with U.S. Census data and with cell phoneâderived estimates for regional travel. The following questions were explored: ⢠How can cell phone data best be used to derive travel estimates and support modeling analyses? Is a behavioral approach to modeling still important and relevant, or can cell phone estimates of travel replace the traditional approach? ⢠Which aspects of the behavioral approach to travel demand modeling can be supported by cell phone data? Which elements of model estimation and model validation can benefit the most from these data? ⢠Is the existing paradigm of travel behavior analysis applicable to this new stream of data? Are new methods required to harvest the value of cell phone data? Are current tools adequate to analyze the new stream of data? ⢠Do cell phone data affect how survey data collection is approached? Does this new technology offer an alternative standalone option for data collection, or can it augment existing methods of data collection? ⢠What are the expansion methods used to arrive at a representative sample? How do these methods compare with the industryâs state of the practice, and what information is needed to assess the representativeness of the sample? ⢠Are the assumptions needed to analyze this new stream of data understood? What biases need to be considered, and how should they be addressed? What aspects of the âblack boxâ need to be better understood? As traditional data and models are contrasted with travel estimates from cell phone data, this report will address the trade-offs between sample sizes, the inferences and assumptions that need to be made, and the richness of information about travel and socioeconomics that characterizes each method.
Summary 3 The Research Approach and the Comparisons Transportation agencies have either purchased or evaluated the purchase of products on the basis of anonymized and aggregate cell phone data in order to supplement, enhance, or replace traditional data. Practitioners need to be clear about the terminology used to describe and analyze cell phone call detail records (CDRs) and to understand the methods used to translate locational data into estimates of travel flows. Key to this discussion is a better under- standing of how cell phone CDR data and other forms of big data compare with traditional forms of data and models that are commonly available to transportation planners. The policy needs that planning agencies face will determine how cell phone data may be used to support these analyses. Interviews of agency staff are summarized to document and understand these policy needs. A checklist has been developed for planners to use when thinking about the uses of CDR data, data strengths and weaknesses, and plannersâ efforts to understand products that may appear as black boxes. The potential utility of cell phone data focuses on the data as a source of travel information, as an input for estimating travel demand matrices, and as a substitute for different modeling components. The research approach used to shed light into the black box was based on a case study from Boston that allowed the research team to make multiway comparisons by using simi- lar measures of travel from CDR data, traditional household surveys, and regional model results. These transparent multiway comparisons were used to highlight the strengths and weaknesses of each source of data. The difficulty of determining, with certainty, what constitutes ground truth estimates is also discussed. The research discussed in the case study is unique, in that it used raw CDR data to develop estimates of activities, their location, and the time of day when they happened. These esti- mates are rolled up to the regional level to produce metrics readily comparable to traditional estimates of travel developed from household surveys and used in regional models. The transparency of the research assumptions, the open discussion of the strengths and weak- nesses of the underlying data and methods, and the step-by-step methodological insights are useful in understanding the properties of similar products in todayâs marketplace. The results are presented in detail and followed by a brief technical summary of key points in each chapter. The analysis of CDR data suggests their potential for (a) supporting fre- quent and targeted data collection of travel patterns for external, long-distance, and special events and visitor travel; (b) identifying seasonal and year-to-year variations in travel; and (c) providing a means of assessing trends in regional travel. Although CDR data cannot be used in traditional travel demand models, these data can be used as an additional source Concept of the Black Box The complexity of travel demand models can lead the public and decision makers to think of models as a black box whose internal structure and underlying assumptions are not known or clearly understood. The modeling community has made efforts over the years to better document, explain, and communicate data, methodologies, and assumptions to decision makers and to the public. A similar level of transparency is critical when new methodologies and assumptions are used to analyze cell phone data to support planning decisions and build travel demand models.
4 Cell Phone Location Data for Travel Behavior Analysis for model validation, to help drive model updates for intermediate years, and to support analyses for model components (e.g., long-distance travel, special generators, visitor travel, special events, and, potentially, corridor studies). Guidelines This report concludes with a chapter on the issues practitioners typically consider about data and models in their daily work. The research team reframes these considerations for CDR locational data to help agency staff assess the potential value of these data. The prac- titioner guidelines are grouped into three categories: administrative considerations, data considerations that staff face, and modeling-related issues that practitioners need to address as they assess the potential value of CDR data. In an epilogue, the research team recommends that practitioners apply the following principles as they evaluate their policy needs, data options, and modeling tools: ⢠Be aware of the assumptions made in processing cell phone data to determine locations and infer activities and purposes. ⢠Recognize that results from traditional surveys and models are built on different sets of assumptions and that ground truth is difficult to establish. ⢠Expect that increases in the quantity of CDR data, improvements in signal and CDR data quality, and the use of machine-learning algorithms are likely to improve methods for analyzing locational data and inferring travel patterns. ⢠Appreciate the uncertainty underlying CDR estimates and traditional data and measures of travel patterns. ⢠Use the conceptual framework based on the behavioral paradigm that examines individu- alsâ travel behavior as a guide. The research indicates that, together, collective industry experience, academic research over the years, and a collaborative approach linking research to practice have helped refine the data design process, spawn new and more sophisticated analytical methods, and increase understanding of travel behavior and its drivers. As new data and methods are introduced, the interpretation of their value and uses through a behavioral framework lens will help improve the state of the art in travel demand forecasting. Major opportunities will likely emerge to harness new ideas, data, and methods to shape innovative practices that have long-term potential to benefit the transportation research community.