Smart data for greener transportation - potentials and pitfalls in evaluating behavioral changes

The SmartChange project studied how new sources of data can enable easier detection of changes in travel behavior. Traditionally, these changes have been captured through surveys - a time consuming, inaccurate and costly data collection procedure. Falling response rates have also undermined the trustworthiness of this type of data. It is harder to reach respondents, and people are less willing to actively respond to surveys. The amount of passively collected data, however, has increased remarkably over the last years. Mobility data is e.g. registered by sensors in the road network or public transport system, and by our extensive use of smart phones. The data volume rapidly grows - and this calls for different tools and methods than survey data. Big data from sensors and the internet of things have a different spatial and temporal granularity than survey data, which opens new opportunities and challenges. During the project period, several measures designed to stimulate a more sustainable transport sector were introduced. To be able to implement efficient policy measures, it is important to assess the impact of the changes. New technology and new sources of data enable new types of analyses in a complex and dynamic transportation system. However, the new sources of data bring new challenges and sources of error to the table. The project investigated both the opportunities and challenges by comparing traditional and new data. Data of mobility patterns and data used for other purposes than they initially were collected for, raise both legal and ethical questions. Since the project started in January 2017, researchers at TØI have carried out a literature study on the topics of smart cities, big data, data protection laws and ethical issues. In addition, there has been a dialogue with the Norwegian Data Protection Authority and The National Committee for Research Ethics in the Social Sciences and the Humanities (NESH). In cooperation with the University of Tartu, Estonia a survey designed to measure peoples' attitudes to the use of mobile phone data was be sent to respondents in Oslo and Tallinn. Results from the survey suggest that there is a certain acceptance among the population in Oslo and Tallinn for such data to be used to combat crime and terror and to improve transport. However, more than half of the respondents believes that mobility tracking via the mobile phone on a general basis threatens to a high or very high degree the protection of privacy. In June 2017, SmartChange organized a symposium called ?Big Data in Transportation Research?. This event brought together international and national transport researchers, as well as people from Norwegian transport agencies and authorities. Several researchers at the symposium raised privacy of the respondents as one of the major challenges to further apply big data in transportation research. Another take home massage from the symposium is that more data is not necessary better, it is the quality of the data that is crucial. In September 2018, SmartChange organized a workshop on smartphone-based travel behavior data collection apps. An objective of the project has been to increase focus and stimulate collaboration on big data within the transport sector. The output from the workshop has been communicated to stakeholders involved in developing the future national travel survey in Norway through presentations by the project members. Preliminary results from the paper «Bayesian spatiotemporal modelling of automated vehicle location data» were presented at the Spatial Statistics 2019 conference in Spain. The paper presented how automated vehicle location data may be applied to detect delays in traffic. This method is especially useful on road sections where other data sources are not available. During summer 2019, an internship student from the University of Harvard participated in the project. He organized data from loop counters for vehicles in Oslo. Further analysis of the data was conducted in a master thesis by another student in 2020. The analysis suggests that traffic in Oslo was reduced due to the increase in road toll fees in 2017. This holds even when comparing the traffic with the traffic in Drammen by difference-in-difference methods. To further investigate the effect of congestion charges in Oslo and experiment with new data sources, the project applied data from the insurance company Fremtid. In this paper, we analyzed a panel data set of 4011 anonymized car owners in the central eastern part of Norway that are tracked by a smart car insurance application. We observe where they live (on neighborhood-level) and how much they drive their car every hour of every day. We combined this with data on tolling costs between where these car owners live and the Oslo city center. Among our findings are that for most of the sample, driving during peak hours and off-peak hours are complimentary goods.

The SmartChange project has contributed to share insight into the limitations and strengths of new data sources available for transportation research. The project has collaborated with other research projects on data from insurance companies, public transport and mobile apps. The findings are promising with regard implementing these data sources in future research projects. Project members have participated in debates about the future national travel surveys in Norway. Despite the increasing amount of new data sources, the members have pointed out that there are no data sources available that may replace the current survey fully. In summary, there are several data sources available that to a limited degree have been applied for transportation research in Norway. However, many data sources that are widely debated among stakeholders, are only to a limited degree available, including e.g. network data from mobile phones, due to data protection legislation.

Recent technological developments have opened new possibilities for more efficient policy interventions and policy evaluations in the transport sector. Traditionally, such interventions, e.g. policies to reduce high levels of local pollutants, have primarily been designed and evaluated through conventional surveys. However, this approach is time consuming and expensive. Furthermore, declining response rates question the surveys' validity and reliability. Big Data, generated through the internet of things and automated real-time monitoring of traffic, potentially offer faster (real-time), more efficient and more representative data for monitoring events and policy interventions in an integrated transport system. This project will develop methods and tools that allow utilizing Big Data for policy interventions and policy evaluations that will be vital for change in the transport system. We selected two case studies to test methods and tools developed in the project: the planned introduction of a congestion pricing scheme and increased parking restrictions in Oslo. The project will have special focus on ethical issues related to the application of Big Data, as Big Data by definition are not given by informed consent. An important outcome of the project will be an understanding of the impacts of policy interventions on all modes of travel and the change to a sustainable transport system. The project will analyze CDR-data (call detail record), automated traffic counts of various travel modes (cars, bicycles and public transport) at multiple locations and conventional travel surveys. TØI has gathered and organized several of these data sources in the BYTRANS project. Furthermore, TØI is taking part in conducting the national travel survey for Norway. The unique opportunity for the combination of these data sources forms an ideal ground to experiment with different data sources, compare them and develop methods that will lead to new knowledge.