Office employees are working from home, schools are closed, and large gatherings with friends and family are discouraged per social distancing recommendations. As we stay at home, we are reminded of how our community interactions have changed with COVID-19. Several reports and maps display how we are practicing social distancing; those numbers and colors illustrate the variation in the number of trips to places such as restaurants, grocery stores, retail stores, parks, gyms, and recreational facilities.
Numerous transportation studies, such as proposed development projects, signal timing adjustments, public roadway improvements, and traffic forecasting require traffic counts. Right now, travel behavior has changed, the number of vehicles on the roads is lower, and the peak hours may not be the same. Historical traffic counts for a specific study location are rarely available or are usually outdated. How can traffic engineers keep important projects moving forward without being able to collect reliable traffic counts manually?
Big data seems to be a good alternative for data collection during these uncertain times.
What is Big Data?
According to the USDOT Intelligent Transportation Systems Joint Program Office, “Big Data is a process of knowledge generation that features the following approaches:
- Data Capture that includes massive datasets encompassing all or most of the population being studied (as opposed to small samples).
- Data Management that features storage in decentralized and virtual locations (i.e., the cloud) and handles both structured and unstructured data.
- Data Analysis that is often automated, with computers doing more of the work to find complex patterns among a large number of variables.”
How is Big Data collected?
The most common sources of Big Data collection for transportation studies are smartphones and connected vehicles. Anonymous location data is aggregated and processed through software that allocates the data into a network that includes roads, bike lanes, and sidewalks. In summary, the aggregated location data provides information on how vehicles, bicycles, and pedestrians move within the network.
What metrics can be provided with Big Data?
Several metrics can be obtained when location data is aggregated. Annual Average Daily Traffic (AADT), trip origin-destination, trip length, trip route, trip purpose, speed, travel time, and turning movement counts (TMCs) are some of the information that can be obtained with Big Data processing.
Why is Big Data a good alternative for traffic counts during COVID-19?
Big Data allows traffic counts to be obtained retroactively. If you want to request data from January, February, or March 2020, before COVID-19 changed travel patterns, Big Data can provide that information.
Is Big Data reliable?
To ensure Big Data for transportation studies is accurate, data providers conduct validation of the metrics. Public sources such as historical permanent counters, household surveys, license plates, and the Census are used for data validation and adjustments as needed. Big Data companies have validated their metrics in the U.S. with a coefficient of determination (R2) of 0.90 or higher.
Does Sain have experience working with Big Data?
Sain staff has worked with Big Data on projects in several states even before COVID-19. Now, as local jurisdictions start to see how Big Data can be an alternative for traffic counts during unusual conditions, Sain staff is ready to apply their knowledge and experience to help local projects move forward.
Dave Duncan, P.E. from the Tennessee Department of Transportation (TDOT) Strategic Transportation Investments Division, Project Coordination & Investigation Office explains why the agency is ready to use Big Data as an alternative to traffic counts:
“As a result of the unusual circumstances with the COVID-19 pandemic and social distancing practices and procedures implemented to curb the spread of the virus, a major concern for TDOT has been the significant drop in traffic volumes. TDOT collects manual and machine counted traffic data on TN roadways regularly for planning activities; however, traffic data collection during this pandemic would provide an unreliable data set to analyze the efficacy of any future investments. Using irregular traffic data can lead to bad decisions and project delays; therefore, TDOT has been investigating different uses for leveraging services that use historical location-based data from anonymized cell phone data and navigation devices in connected cars and trucks, and Internet of Things (IoT) sensors for microsimulations developed to analyze project alternatives. This data has been collected for years across the entire roadway network from different vendors which are capable of algorithmically transforming the data into normalized traffic patterns.”
Is there a future for Big Data applications to traffic studies after COVID-19?
Big Data has been applied to transportation studies before COVID-19, and it seems to be an excellent alternative to manual traffic counts even when traffic conditions start to stabilize.
- Why use Big Data for transportation studies
in the future?
- Availability to look at data over the entire year: the typical manual data collection of TMCs only includes observations on a specific day for a few hours in the morning and a few hours in the evening. The collected data is assumed to be representative of the entire year, which has proven to be incorrect in hundreds of studies across the country. One of the major benefits of Big Data is the availability to look at data over an entire year, which makes it easier to capture situations such as an incident upstream of a study intersection or a parallel corridor which impacts your study corridor. Manual TMCs would not capture this but would still be used for analysis and forecasting purposes.
- Data collection without field deployments or within a data collection window: Big Data can be easily obtained from the software without field deployments. Further, data can be obtained retroactively as opposed to scheduling future data collection.
- Weekends and Weekday off-peak conditions: historical counts are rarely available for off-peak times on weekdays or during the weekends. Big Data allows data aggregation by any hour of a day and any day of the week.
- What are some of the disadvantages of using
Big Data for transportation studies?
- Detailed information is not always available: some traffic studies require detailed information such as heavy vehicle percentages. Also, Big Data typically provides counts for 1-hour increments instead of 15-minute increments. Manual traffic counts are still recommended when very detailed information is needed.
- Costs can be higher than other traffic count methods: depending on the project, costs can be higher than traditional manual traffic counts. This issue is usually a concern for small projects.
- Data may not be as accurate: at small intersections, for example, where permanent counters may not be in place, data may not be accurate as the validation process needs other sources of data.
In summary, Big Data seems to be a good data collection alternative for transportation studies during COVID-19 as well as after traffic conditions become stable again. It is important to note that every source of data collection is likely to have errors. Each jurisdiction should review and validate multiple data sources based on their local conditions before using any dataset.