Effectiveness of Disseminating Traveler Information on Travel Time Reliability: Implement Plan and Survey Results Report
CHAPTER 5. TRAVEL TIME RELIABILITY DATA MANIPULATION
Each region participating in the study provided historical traffic datasets as a source for determining the Travel Time Reliability (TTR) calculations utilized by the mobile application, website, and 511 systems. To ensure compatibility across these platforms and provide information that could easily be understood by the study participants, the study team developed the following requirements for each dataset:
- Average segment-based travel time data with origins and destinations corresponding to the majority of entry/exit points along each of the study corridors by direction. These average values are used as the "typical" travel time for display in the traveler information applications.
- For each of the segments in the study corridors and for each aggregation period, the 95th percentile travel time for use in determining the worst-case travel times.
- Travel time data aggregated by day of week in at least hourly intervals for a 6-month period or more. The aggregation time (e.g., 15 minutes, hourly) limits the resolution of the departure and arrival times in the traveler information applications, and the 6-month timeframe allows for an average traffic flow to be determined without being influenced by singular extraordinary traffic conditions on specific days that can occur during a shorter timeframe. The 6-month timeframe leads to a larger sample size, reduced variability of the data, and has a positive impact on the quality of the data. (Although the study team did not perform exhaustive checks on the quality of data, it is recommended that metrics such as sample size and data variability be provided in the future to make these assessments possible).
- The most recent historical dataset possible in order to reflect the latest typical traffic conditions as accurately as possible.
Based on the availability of data in each region, datasets were provided from different sources, timeframes, and formats. In order to present the data on the mobile application, website, and 511 system in a consistent format, the data from each region had to be manipulated into a common data format. The data sources and required manipulation specific to each region are described in the following sections. The final result of the data manipulation was a comma-delimited file for each region for each day of the week containing location identification information, aggregation start time, average travel time, and 95th percentile travel time. These files were used as the data source for the App, website, and 511 systems. An example of the processed dataset for the Triangle region is provided in Table 4.
| Roadway Name | Direction of Travel | Origin Cross Street | Destination Cross Street | Segment Length (mi) | Summary Time | Travel Time (min) | 95th Perc. Travel Time (min) | Speed (mph) | 95th Perc. Speed (mph) |
|---|---|---|---|---|---|---|---|---|---|
| I-40 | Eastbound | US-15/ US-501 | NC-54 | 2.94933 | 1/1/2013 0:00 | 160 | 180 | 66.36 | 58.99 |
| I-40 | Eastbound | US-15/ US-501 | NC-54 | 2.9493 | 1/1/2013 0:15 | 155 | 172 | 68.50 | 61.73 |
| I-40 | Eastbound | US-15/ US-501 | NC-54 | 2.9493 | 1/1/2013 0:30 | 155 | 162 | 68.50 | 65.54 |
| I-40 | Eastbound | US-15/ US-501 | NC-54 | 2.9493 | 1/1/2013 0:45 | 159 | 167 | 66.78 | 63.58 |
WEST HOUSTON AND NORTH HOUSTON TRANSPORTATION STUDY
The Houston region has an extensive deployment of ITS-based sensors installed throughout each of the study corridors that provide speed and travel time information through the region's traffic management center, Houston TranStar®. The source of the Houston study data was the information collected by these sensors that utilize either Bluetooth or toll-tag-based re-identification for estimating travel times. The sensors are operated by the Texas Department of Transportation (TxDOT), which provided data for the study's usage for the I-10 Katy, I-10 Katy Managed Lanes, Westpark Tollway, I-45 North, I-45 North HOV, and Hardy Toll Road corridors.
The origins and destinations for each travel time segment were based on the locations of the roadside sensors. In most cases, the sensors were located near major entry and exit points along the corridors, with 1- to 3-mile spacing. The software internal to Houston TranStar® collects and processes the travel time data in both real-time and historically, and aggregates the data into 15-minute summaries by day of week. For the purpose of this study, a historical dataset for July through December of 2014 was used.
For each 15-minute period and for each day of the week, the dataset contained a location identifier (including the roadway name, direction of travel, origin cross street, and destination cross street), a timestamp indicating the time of the summary, an average travel time, and a 95th percentile travel time. Because the Houston dataset was the first to be manipulated and met all of the original requirements, this data format was used for the other two study regions.
The Houston data contained additional background information on the process that generated the historical data which allowed the study team to determine data quality. In producing the historical data, quality was assessed by using the number of sample periods with available data along with the standard deviation of the individual speed samples in miles per hour. For a historical average to be viable, 80% of the time periods had to contain data samples and the standard deviation of the speed samples had to be less than 15 mph for each segment during each time period for each individual day. For example, in a 6-month period using 15-minute summaries, 80% of the Mondays at 7:00 AM needed to have samples. Additionally, for each time period on each day, the segment had to have a standard deviation that was less than 15 mph. There were no instances where the data did not meet the quality control check.
NORTH COLUMBUS TRANSPORTATION STUDY
The North Columbus study area on I-71 used INRIX data provided by the Ohio Department of Transportation (ODOT). INRIX collects data from Global Positioning System (GPS) equipped mobile devices to estimate traffic conditions. The dataset provided by ODOT contained aggregated hourly averages of travel time and 95th percentile travel time by day of week for July through December of 2014. Although 15-minute aggregation data was preferred because of the lower time resolution it provides, those data could be obtained only for a short timeframe consisting of several weeks. To limit the historical averages being influenced by extraordinary traffic conditions on specific days, the study team decided that having hourly aggregations over a 6-month timeframe was preferable to having 15-minute aggregations over a much shorter timeframe. For the contents of the dataset, INRIX was responsible for providing the travel time, and ODOT calculated a reliability index that was translated into the 95th percentile travel time. The INRIX data was pre-processed and screened for quality by ODOT and did not contain any quality assessment metrics in the output used by the study team.
The travel times associated with INRIX data are provided for predefined segments called Traffic Message Channels (TMCs). The TMC segments typically correspond to very short sections of roadway and provide more granular information than was needed for the study. In order to provide travel time information between exit and entry points on the corridor, data from multiple TMC segments had to be aggregated together. For instance, there might be five or more TMC segments between an entry and exit point on the freeway. In this case, data from all the TMC segments corresponding to that section of roadway were added together to obtain a total travel time. An example of the TMC segments used in the North Columbus dataset are provided in Table 5. The example shows an origin and destination on I-71 between I-270 and OH-161 that comprises two TMC segments, which are uniquely identified by the TMC ID.
| TMC ID | Roadway | Direction | Origin | Destination | Length (mi) | Average Travel Time (min) | 95th Perc. Travel Time (min) |
|---|---|---|---|---|---|---|---|
| 122-04201 | I-71 | Southbound | I-270/Exit 119 | OH-161/ Exit 117 | 0.57308 | 0.54 | 0.59 |
| 122N04201 | I-71 | Southbound | I-270/Exit 119 | OH-161/ Exit 117 | 1.275554 | 1.21 | 1.43 |
TRIANGLE TRANSPORTATION STUDY
The Triangle study area along the I-40 corridor in the Raleigh-Durham-Chapel Hill region used data provided by the Durham-Chapel Hill-Carrborro Metropolitan Planning Organization (DCHC MPO), which obtained them from HERE. HERE is a traffic data provider similar to INRIX that collects data from GPS-equipped probes to estimate traffic conditions. The dataset provided by the DCHC MPO was 15-minute data summarized by day for the entire calendar year of 2013. The daily summarized data included a location identifier, segment length, average speed, and various speed percentile values. For the purpose of this study, only the 95th percentile value was used. To make the dataset compatible with the required format, the study team had to aggregate the data into day of week summaries and convert the speed values provided into travel time by using the segment length value.
Similar to INRIX, the speeds associated with the HERE data are provided for predefined TMC segments that had to be aggregated in order to provide a total travel time between entry and exit points along the corridor.
The HERE data contained the number of samples and standard deviation for each time period, and these metrics were used to assess data quality using thresholds similar to the ones applied to the Houston data. All of the variability and sample size issues were addressed by the one-year timeframe that the data represented.
SUMMARY OF STUDY TRAVEL TIME RELIABILITY DATASETS
The historical TTR datasets for each study location that were used in the study are summarized in Table 6.
| Study Location | Data Source | Data Timeframe | Aggregation Time |
|---|---|---|---|
| West/North Houston (Texas) | TxDOT roadside sensors (Toll tag/Bluetooth) | July-December 2014 | 15 minutes by day of week |
| North Columbus (Ohio) | INRIX provided by ODOT | July-December 2014 | 1 hour by day of week |
| Triangle (North Carolina) | HERE provided by DCHC MPO | January-December 2013 | 15 minutes by day of week |
WEBSITE DATA INTERFACE TECHNICAL DESCRIPTION
The backend architecture for the website provided a data interface framework for each of the information channels used in Phase 2 of the study. A web Application Programming Interface (API) was developed that allowed other applications to query the historical datasets that were developed for each study region. Both the traditional website and the 511 system made queries to the web interface to obtain the traffic conditions data. The mobile application embedded the traditional website into its framework, thus using the same data source.
To initiate a query to the web service, the client (i.e., website, 511 system) made a call to a web address with the following parameters included:
- Starting Location ID.
- Ending Location ID.
- Time of Day.
- Date of Travel.
- Departure or Arrival Calculation.
- Assembly Type.
Based on the information passed via the parameters, the web service queried the appropriate historical dataset and returned a text string containing the approximate travel time, buffer time, and predicted arrival or departure time. The different information channels then were able to relay this information in an appropriate format (i.e., webpage via the website and mobile application, via 511). A diagram of the web service architecture is shown in Figure 24.
Figure 24. Graphic. Web service architecture.
