Texas Transportation Institute The Texas A&M University System 3135 TAMU College Station, TX 77843-3135 979-845-8829 Fax: 979-845-6008 http://tti.tamu.edu shawn-turner@tamu.edu March 7, 2016 TECHNICAL MEMORANDUM TO: Paul Czech, Minnesota Department of Transportation FROM: Shawn Turner, Texas Transportation Institute (TTI) SUBJECT: TTI Review of Selected Inrix Data in Minneapolis-St. Paul This memo documents the key findings of a TTI review of Inrix data for selected roadway segments in the Minneapolis-St. Paul area. The key findings and conclusions are as follows: Comparability of Inrix Data to Other Data Sources TTI review indicated that the Inrix speed data followed expected temporal and day-ofweek patterns and trends. Inrix speed data during the overnight hours was highly variable (due to much smaller sample sizes) and would not be suitable for use by Mn/DOT. However, overnight data (10 pm to 6 am) is typically not used in congestion analyses. Comparison of Inrix speed data to the Mn/DOT fixed-point sensor data indicated that Inrix freeway speeds were consistently slower than Mn/DOT speeds. For a two-year average speed comparison, the mean absolute difference between average speeds was about 5 mph during light traffic and 10 to 15 mph during the peak periods. It is important to recognize that, at this point, we cannot say that the Mn/DOT data is most accurate and all of this difference is due to inaccuracies in Inrix data. These differences are most likely caused by: 1) different vehicle mix (2006-2007 Inrix data is mostly from commercial fleet vehicles); and 2) different speed measurement techniques. The most recent Inrix freeway speed data (January-February 2008) compared more favorably to Mn/DOT sensor data, as the mean absolute difference was between 5 and 10 mph during the peak periods. Comparison of Inrix arterial street speed data to Mn/DOT signal retiming data indicated that these data were in general agreement, although no differences were quantified. There were several methodological differences that prevent this comparison from being conclusive: 1) the Mn/DOT data was collected over a few days, whereas the Inrix data was collected over the entire month; 2) the Mn/DOT data consistently had fewer samples than the Inrix data; and, 3) data aggregation in the Inrix data prevented a true “apples-toapples” comparison in the temporal and spatial dimensions. Mobility Analysis Program Inrix Data Sample Sizes TTI review indicated that, on most of the routes selected by Mn/DOT, the Inrix data sample sizes were greater than the sample sizes typically gathered through public agency travel time data collection. The current mix of vehicles providing speed data to Inrix appears to over-represent fleet vehicles (such as large trucks) and under-represents passenger vehicles as they occur in the traffic stream. The differences in speed and operating characteristics of fleet and passenger vehicles are more distinct in stop-and-go traffic congestion, as was apparent in several of the comparisons in this memo. The current mix of vehicles providing speed data to Inrix leads to the highest proportion of samples during typical workday hours, with much lower sample proportions during the weekend and off-peak hours. However, these lower sample proportions are less critical during these times since most congestion analyses are focused on the weekday peak periods. The number of Inrix data samples has significantly increased in the past two years. As GPS is becoming more common to track vehicle locations, Inrix and other real-time traffic information companies are aggressively pursuing these new probe vehicle data sources. Suitable Applications for this Data Based upon the Inrix speed data and associated sample sizes for the Mn/DOT-selected routes, we conclude that the most favorable applications for this type of data are those that require historical data on an aggregate basis (e.g., a monthly or annual average, a corridor-wide performance measure, etc.). We also conclude that the most favorable applications at this time are those that require relative accuracy (i.e., making comparisons among routes and over time). Examples of possible applications include but are not limited to: Identifying most congested routes or segments for further study and improvement; Ranking or prioritizing routes for funding based on average congestion levels; Monitoring and identifying routes with the highest increase in average congestion levels (on monthly or annual basis); Using as a general indicator in before-after studies of mobility improvements; and, Developing performance measures on an aggregate basis. There are certain applications that TTI cannot recommend for this type of data at this time. These applications typically require larger sample sizes, more disaggregate data (on a daily basis), or absolute accuracy. Development of reliability measures that capture day-to-day variation; Detailed signal retiming and synchronization studies; Before-after studies that require detailed data for specific time periods and locations; and, Real-time traffic information on certain lower functional roadway classes (those with small sample sizes). 2 There are two additional conclusions that we have developed: Private sector companies currently only collect and provide travel speeds, so performance measures that require traffic volumes or flow rates will require this information from another source, such as statewide or metro area traffic counts. Certain performance measures do not require traffic volumes (e.g., travel time index, congestion duration, percent of roadway miles with peak period congestion, buffer and planning time index) or only require traffic volumes to weight performance measures in aggregated subarea or areawide measures. Other performance measures, such as various delay measures and throughput, do require more accurate traffic volumes. It appears that there will be a market for the private sector provision of travel speed data for various traveler information and telematics applications. From this, we conclude that historical private sector data should be available to public agencies in the near term (at some cost). However, the current companies in this market may change or be consolidated as this industry grows. Additionally, the sample sizes behind private sector data are predicted to increase dramatically in the next few years as GPS-enabled devices proliferate in mobile phones and in-vehicle navigation devices. The caution associated with this sample size increase is that before-after studies will need to carefully consider the effects of smaller samples sizes on the accuracy of the “before” data as compared to the “after” data. The rest of this memo provides more details of the TTI review to support these findings and conclusions. INTRODUCTION In May 2008, the Texas Transportation Institute (TTI) contracted with Inrix, Inc. to purchase a sample of travel speed data on selected arterial street and freeway segments in the MinneapolisSt. Paul area. The data were purchased on behalf of the Minnesota Department of Transportation ( Mn/DOT) with the objective of evaluating the quality of Inrix data for mobility and reliability monitoring and reporting purposes. The evaluation was conducted using three strategies: 1. Visual checks of basic traffic patterns and trends; 2. Comparison with Mn/DOT freeway fixed-point sensors; and, 3. Comparison with Mn/DOT test vehicle data collected for signal retiming purposes. BACKGROUND Routes Mn/DOT planning staff provided a preferred list of routes that were of primary interest. Inrix staff matched this against their data source to ensure sufficient samples (based on their criteria). As Figure 1 and Table 1 indicate, Inrix was able to provide data on nearly all of the roadways 3 requested by Mn/DOT. Table 1 also indicates that nearly all of the evaluated routes were based on Mn/DOT priority and not the availability of adequate Inrix data samples. Priority 1 2 n.a. Table 1. Routes Requested TTI & Mn/DOT and Provided by Inrix Route & Area Requested Provided by Inrix MN 65: North MN 65: I-694 to HWY 242 US 61: Northeast US 61: HWY 36 to HWY 96 US 61: Southeast US 10/US 61: I-94 to I-494 MN 7: Southwest MN 7: MN 41 to US 169 MN 55: West MN 55: HWY 50/HWY 10 to I-494 US 169: North US 169: HWY 130 to US 10 US 10: North US 10/US 169: US 169 to US 169/HWY 101 MN 36: East Substituted CO HWY 14: HWY 65 to I-35W HWY 13: South HWY 13: HWY 101 to HWY 77 HWY 5: Southwest Not provided I-394: West Substituted I-494: US 212 to HWY 77 Figure 1. Routes on Which Inrix Provided Travel Speed Data US 10/US 169 CO HWY 14 US 169 MN 55 MN 65 US 61 MN 7 US 10/US 61 I-494 HWY 13 4 Inrix Data Inrix provided data from January 2006 through February 2008 in three distinct data sets: 1. 60-minute average speeds for each day of the week and month of the year o Location (TMC code) o Day-of-week/time (168 bins) for each month of each year o Average speed o Number of data samples 2. 15-minute average speeds for each day of the week o Location (TMC code) o Day-of-week/time (672 bins) – all months combined into a single average o Average speed o Number of data samples 3. 15-minute “reliability” statistics for each day of the week o Location (TMC code) o Day-of-week/time (672 bins) o Average speed o Speed percentiles: 10, 15, 25, 50, 85 o “Reliability” – frequency of time/samples below 30, 50, and 60 mph Inrix provided a license to TTI to use this data for evaluation purposes and also granted TTI permission to redistribute aggregate or derivative summary statistics from the speed data. The redistribution of the original speed data provided to TTI, as well as the number of data samples, was prohibited by the Inrix/TTI licensing agreement. Inrix also provided TTI with a location code table that uses the RDS-TMC (Radio Data System – Traffic Message Channel) location referencing system. This location referencing system originated in Europe and is a de facto standard in the United States used by many of the private traffic information companies. In North America, the TMC location codes table is maintained and regularly updated by NAVTEQ and Tele Atlas, two of the dominant map database companies in the market. TMC location codes are unique numeric sequential identifiers (typically 4 to 5 digits) assigned to intersections and interchanges that are significant for traffic messages and routing purposes. An internal path (designated by N or P) is the directional road segment that is internal to the identified intersection or interchange. An external path (designated by - or +) is the directional road segment that leads up to the intersection or interchange (using local driving conventions). A simple lookup table allows the TMC code and path to be cross-referenced to route identifier (e.g., MN 55, US 169) and cross street limits. The different colored circles shown in Figure 1 indicate the TMC code locations for the routes on which Inrix provided historical travel speed data. Figure 2 shows a zoomed view for one route, MN 7 (part of the I-494 route also can be seen in the bottom right corner). As the figure 5 indicates, the selection and spacing of TMC code locations varies based on the road network. The assignment of TMC codes to specific intersections or interchanges is a decision made by NAVTEQ and Tele Atlas based on their significance for traffic messages and routing purposes. The typical length of roadway (i.e., TMC path) for which travel speed data was provided ranged from 0.1 to 4.75 miles, with an average length of 1.1 miles. For parts of the evaluation, these original link definitions were used to facilitate comparisons with Mn/DOT freeway sensor data. In other parts of the evaluation, these original links were combined to produce longer roadway segments that ranged from 1.5 to 9.7 miles (see Table 2). Summary travel speed statistics were then calculated for these longer roadway segments and are shown in later parts of this report. Figure 2. Example of TMC Code Locations on MN 7 6 Table 2. Combined Roadway Segments Used in Some Parts of the Evaluation Route and Limits MN 65 NB: I-694 to US 10 MN 65 NB: US 10 to HWY 242 MN 65 SB: HWY 242 to US 10 MN 65 SB: US 10 to I-694 US 61 NB: HWY 36 to I-694 US 61 NB: I-694 to HWY 96 US 61 SB: HWY 96 to I-694 US 61 SB: I-694 to HWY 36 US 10/US 61 EB: I-94 to I-494 US 10/US 61 WB: I-494 to I-94 MN 7 EB: MN 41 to I-494 MN 7 EB: I-494 to US 169 MN 7 WB: US 169 to I-494 MN 7 WB: I-494 to MN 41 HWY 55 EB: HWY 50/HWY 10 (Rockford) to CR 116 HWY 55 EB: CR 116 to I-494 HWY 55 WB: I-494 to CR 116 HWY 55 WB: CR 116 to HWY 50/HWY 10 (Rockford) US 169 NB: HWY 130 to HWY 610 US 169 NB: HWY 610 to US 10 US 169 SB: US 10 to HWY 610 US 169 SB: HWY 610 to HWY 130 US 10/US 169 EB: 169/HWY 101 to CR-56 US 10/US 169 EB: CR-56 to US 169 US 10/US 169 WB: US 169 to CR-56 US 10/US 169 WB: CR-56 to US 169/HWY 101 HWY 13 NB: HWY 101 to I-35W HWY 13 NB: I-35W to HWY 77 HWY 13 SB: HWY 77 to I-35W HWY 13 SB: I-35W to HWY 101 I-494 EB: US 212 to US 169 I-494 EB: US 169 to HWY 100 I-494 EB: HWY 100 to I-35W I-494 EB: I-35W to HWY 77 I-494 WB: HWY 77 to I-35W I-494 WB: I-35W to HWY 100 I-494 WB: HWY 100 to US 169 I-494 WB: US 169 to US 212 CO HWY 14 EB: HWY 65 to CR-17 CO HWY 14 EB: CR-17 to I-35W CO HWY 14 WB: I-35W to CR-17 CO HWY 14 WB: CR-17 to HWY 65 7 Length (miles) 4.7 4.3 4.3 4.7 1.9 3.2 3.2 1.9 4.8 4.8 6.8 2.6 2.6 6.8 9.7 4.9 4.9 9.7 2.8 5.4 5.4 2.8 6.7 3.5 3.5 6.7 4.5 4.2 4.2 4.5 1.5 2.2 2.6 2.5 2.5 2.6 2.2 1.5 3.4 4.4 3.4 4.4 Speed Limit Range (mph) 40 to 55 55 to 60 55 to 60 40 to 55 50 50 50 50 45 to 60 45 to 60 50 to 55 45 to 55 45 to 55 50 to 55 50 to 55 55 to 55 55 to 55 45 to 55 55 30 to 55 30 to 55 55 60 to 65 60 60 60 to 65 45 to 55 50 to 55 50 to 55 55 60 60 60 60 60 60 60 60 50 to 55 50 to 55 50 to 55 50 to 55 EVALUATION FINDINGS In an ideal evaluation scenario with no budgetary constraints, extensive “ground truth” travel speeds would be obtained on the evaluation routes to determine the error associated with Inrix travel speeds. In this case, the collection of “ground truth” travel times for long periods of time (two years in this case) is simply not feasible. Therefore, TTI researchers assessed the quality of the Inrix data using three strategies: 1. Perform visual checks of basic traffic patterns and trends in Inrix data; 2. Compare Inrix data with Mn/DOT freeway fixed-point sensor data; and, 3. Compare Inrix data with Mn/DOT test vehicle data collected for signal retiming purposes. The findings from these three strategies are described in the following sections. Visual Checks of Traffic Patterns and Trends Basic time-of-day speed profile charts were prepared for the two distinct average speed data sets: 1. 15-minute average speeds (all months combined); and, 2. 60-minute average speeds (month-by-month). Figures 3 and 4 show examples of the 15-minute average speed profiles on which all travel speeds from January 2006 through February 2008 are combined. It should be noted that Inrix provided separate averages for each day of the week. Because the differences between the five weekdays were relatively minor, TTI combined the five weekday values into a single weekday average. Appendix A contains similar 15-minute average speed charts for all 42 roadway segments as defined in Table 2. Figures 5 and 6 show examples of the 60-minute average speed profiles on which all travel speeds were reported on a month-by-month basis (year 2007 shown). It should be noted that Inrix provided separate averages for each day of the week and each month of the year from January 2006 through February 2008. Because the differences between the five weekdays were relatively minor, TTI combined the five weekday values into a single weekday average. Appendix B contains similar 60-minute average speed charts for all 42 roadway segments as defined in Table 2. 8 Figure 3. Example of 15-Minute Average Speed Profile, US 169 NB US 169 NB: HWY 130 to HWY 610 70 15-Minute Average Weekday Speeds, Jan 2006 through Feb 2008 Average Corridor Speed (mph) 60 Posted speed limit is 55 mph 50 40 30 20 10 0 12 AM 2 AM 4 AM 6 AM 8 AM 10 AM 12 PM 2 PM 4 PM 6 PM 8 PM 10 PM Time of Average Weekday Figure 4. Example of 15-Minute Average Speed Profile, US 169 SB US 169 SB: HWY 610 to HWY 130 70 15-Minute Average Weekday Speeds, Jan 2006 through Feb 2008 Average Corridor Speed (mph) 60 Posted speed limit is 55 mph 50 40 30 20 10 0 12 AM 2 AM 4 AM 6 AM 8 AM 10 AM 12 PM 2 PM 4 PM Time of Average Weekday 9 6 PM 8 PM 10 PM Figure 5. Example of 60-Minute Average Monthly Speed Profile for 2007, US 169 NB US 169 NB: HWY 130 to HWY 610 70 Hourly Average Weekday Speeds by Month in 2007 60 Average Corridor Speed (mph) Posted speed limit is 55 mph 50 2007 - 1 2007 - 2 2007 - 3 40 2007 - 4 2007 - 5 2007 - 6 30 2007 - 7 2007 - 8 2007 - 9 20 2007 - 10 2007 - 11 2007 - 12 10 0 12 AM 2 AM 4 AM 6 AM 8 AM 10 AM 12 PM 2 PM 4 PM 6 PM 8 PM 10 PM Time of Average Weekday Figure 6. Example of 60-Minute Average Monthly Speed Profiles for 2007, US 169 SB US 169 SB: HWY 610 to HWY 130 70 Hourly Average Weekday Speeds by Month in 2007 60 Average Corridor Speed (mph) Posted speed limit is 55 mph 50 2007 - 1 2007 - 2 2007 - 3 40 2007 - 4 2007 - 5 2007 - 6 30 2007 - 7 2007 - 8 2007 - 9 20 2007 - 10 2007 - 11 2007 - 12 10 0 12 AM 2 AM 4 AM 6 AM 8 AM 10 AM 12 PM 2 PM 4 PM Time of Average Weekday 10 6 PM 8 PM 10 PM Comparison with Mn/DOT Freeway Fixed-Point Sensor Data The second strategy used by TTI to evaluate the quality of Inrix data was to compare it with speeds collected by Mn/DOT’s freeway fixed-point sensors. The underlying speed measurement technique for each data source is different, as the Mn/DOT sensor is fixed and measures spot speed at one location, whereas the Inrix data are mostly spot speeds collected from various locations near the Mn/DOT fixed-point sensor. Another difference is that the Mn/DOT sensors measure the spot speeds of all vehicles it detects, whereas Inrix data includes a limited sample of vehicles in the traffic stream. Notwithstanding this difference in measurement technique, a comparison was useful to see how similar the data sources were. The comparison was performed at numerous locations along I-494 (Figure 7). Average speeds along the entire I-494 segment were first compared (Figures 8 and 9). Next, individual Mn/DOT sensor locations were compared to the nearest Inrix link data and the error results were aggregated along the entire I-494 segment (Figures 10 and 11). Figure 7. Comparison of Inrix Data to Mn/DOT Fixed-Point Sensors Note: Inrix TMC code locations = small red circles Mn/DOT fixed-point sensors = large blue circles These comparisons indicated that Inrix speeds were typically 5 to 10 mph slower than the Mn/DOT speeds. The differences were greatest during the peak periods, when speeds were the slowest. We hypothesize that most of this difference is due to: 1. Vehicle bias – We believe that Inrix speeds are mostly collected from fleet vehicles (large trucks) that have different operating characteristics than passenger vehicles in traffic congestion. For example, a previous TTI study in Houston showed that truck speeds were about 4% to 8% (2 to 4 mph) slower than passenger vehicle speeds; 2. Different measurement techniques – Mn/DOT sensor speeds are measured at fixed location, whereas Inrix speeds are measured at various random locations on the same link; and, 3. Sampling – Inrix speeds are only collected from a sample of the vehicle traffic, whereas Mn/DOT sensors collect speeds from every vehicle. In comparing the most recently available Inrix data from January and February 2008 (Figure 9), it appears that with the increased data samples in recent months (see later sections of this memo), the differences between Inrix and Mn/DOT speeds have decreased. 11 Figure 8.Average Speed Comparison across Two Years, I-494 Comparison of Average Weekday Speeds along I-494 (15-Minute Intervals, Jan 2006 through Feb 2008) 70 Average Speed (mph) 60 50 40 30 20 10 0 12 AM 2 AM 4 AM 6 AM 8 AM 10 AM 12 PM 2 PM 4 PM 6 PM 8 PM 10 PM Time of Average Weekday MnDOT Detectors Inrix Probes Mean Absolute Difference Figure 9. Average Speed Comparison in Jan-Feb 2008, I-494 Comparison of Average Weekday Speeds along I-494 (60-Minute Intervals, Jan-Feb 2008) 70 Average Speed (mph) 60 50 40 30 20 10 0 12 AM 2 AM 4 AM 6 AM 8 AM 10 AM 12 PM 2 PM 4 PM 6 PM Time of Average Weekday MnDOT Detectors Inrix Probes 12 Mean Absolute Difference 8 PM 10 PM Figure 10. Speed Error at Individual Locations across Two Years, I-494 Distribution of Average Speed Differences along I-494 (15-Minute Intervals, Jan 2006 through Feb 2008) 600 Inrix speeds slower than MnDOT speeds Inrix speeds faster than MnDOT speeds 500 Frequency 400 300 200 100 0 -50 -40 -30 -20 -10 0 10 20 30 40 50 Difference (MnDOT - Inrix) Figure 11. Speed Error at Individual Locations in Jan-Feb 2008, I-494 Distribution of Average Speed Differences along I-494 (60-Minute Intervals, Jan-Feb 2008) 120 Inrix speeds faster than MnDOT speeds Inrix speeds slower than MnDOT speeds 100 Frequency 80 60 40 20 0 -50 -40 -30 -20 -10 0 10 Difference (MnDOT - Inrix) 13 20 30 40 50 Comparison with Mn/DOT Test Vehicle Data The third strategy used by TTI to evaluate the quality of Inrix data was to compare it with “floating car” speeds collected by Mn/DOT contractors for signal retiming purposes. As with the previous comparison, the underlying speed measurement technique for each data source is different, as the Mn/DOT floating car measure space mean speed (average speed over a distance), whereas the Inrix data are mostly spot speeds collected from various locations over a similar road link. Another difference is that the Mn/DOT signal retiming data was collected at limited times during the day, and only on several days. The Inrix data is collected on a routine basis for every day of the year. Unfortunately, the Mn/DOT signal retiming data was the only travel speed data that was readily available on the arterial streets. Figure 12 shows the two arterial streets on which comparisons were made: MN, 55 and MN 65. As with the I-494 segments, speed comparisons were made on an aggregated basis where the link definitions for measured speeds matched. An attempt was also made to match the month of data collection, although this was not possible in every case. Figure 12. Comparison of Inrix Data to Mn/DOT Signal Retiming Data 2 Comparison Corridors MN 55 MN 65 18 Figures 13 through 16 show the results of this comparison. Note that in many time periods, the average speeds from the Mn/DOT were based on fewer speed data samples than the Inrix data. From these comparisons, one is unable to determine which speed data is more accurate, only how the two speed data sources compare to one another. 14 Figure 13. Comparison of Speeds along MN 55 Eastbound Comparison of Average Weekday Speeds along TH 55 EB 70 Hourly Average Weekday Speeds by Month in May 2007 60 Posted speed limit is 55 mph Average Speed (mph) 50 MnDOT Average 40 30 Inrix Monthly Average 20 10 0 12 AM 2 AM 4 AM 6 AM 8 AM 10 AM 12 PM 2 PM 4 PM 6 PM 8 PM 10 PM Time of Average Weekday Figure 14. Comparison of Speeds along MN 55 Westbound Comparison of Average Weekday Speeds along TH 55 WB 70 Hourly Average Weekday Speeds by Month in May 2007 60 Posted speed limit is 55 mph Average Speed (mph) 50 MnDOT Average 40 Inrix Monthly Average 30 20 10 0 12 AM 2 AM 4 AM 6 AM 8 AM 10 AM 12 PM 2 PM Time of Average Weekday 15 4 PM 6 PM 8 PM 10 PM Figure 15. Comparison of Speeds along MN 65 Northbound Comparison of Average Weekday Speeds along MN 65 NB 70 Hourly Average Weekday Speeds 60 Average Speed (mph) 50 MnDOT Average (Aug-Sep 2003) 40 30 Inrix Average (Aug-Sep 2006,2007) 20 10 0 12 AM 2 AM 4 AM 6 AM 8 AM 10 AM 12 PM 2 PM 4 PM 6 PM 8 PM 10 PM Time of Average Weekday Figure 16. Comparison of Speeds along MN 65 Southbound Comparison of Average Weekday Speeds along MN 65 SB 70 Hourly Average Weekday Speeds 60 Average Speed (mph) 50 MnDOT Average (Aug-Sep 2003) 40 30 Inrix Average (Aug-Sep 2006,2007) 20 10 0 12 AM 2 AM 4 AM 6 AM 8 AM 10 AM 12 PM 2 PM Time of Average Weekday 16 4 PM 6 PM 8 PM 10 PM Evaluation of Data Sample Sizes As indicated previously, Inrix did provide number of data samples associated with the average speeds. However, the number of data samples is protected by the Inrix-TTI nondisclosure agreement. Based on our analysis of the confidential information regarding sample sizes, we developed a qualitative assessment for the routes that were evaluated in this effort. The qualitative assessment (see Table 3) is based on these three factors: 1. Absolute number of data samples; 2. Relative number of data samples as compared to current agency practice; and, 3. Comparability of Inrix data to Mn/DOT data and expected patterns/trends. Table 3. Qualitative Assessment Based on Evaluation Criteria Route Evaluation Criteria Absolute samples Relative samples Comparability Evaluated US 10/US 169 1 1 2 HWY 13 1 1 2 I-494 1 2 2 MN 65 2 2 2 US 10/US 61 2 2 2 MN 55 2 2 2 US 169 2 2 2 CO HWY 14 2 2 3 US 61 3 3 2 MN 7 3 3 3 Note: 1=high (most desirable); 2=moderate; 3=low (least desirable) Composite Score 4 4 5 6 6 6 6 7 8 9 At TTI’s request, Inrix did provide permission to TTI to release the general patterns and trends associated with the speed data samples. These are shown in Figures 17 through 22. Of particular note is the increase of Inrix data samples from January 2006 to January 2008. By comparing Figure 19 to Figure 20 (both charts have the same y-axis scale), one can see that the average number of data samples on each route has at least doubled on most routes, and tripled or quadrupled on some routes. An even more pronounced increase in data samples can be seen by comparing Figure 21 to Figure 22 (both charts have the same y-axis scale). In these charts, nearly all of the average weekday samples have quadrupled between 2006 and 2008. 17 Figure 17. Time-of-Day Trends in 15-Minute Data Samples, By Route Total Number of 15-Minute Samples in 26 Months Averaged across all links (both directions) on each corridor 300 Number of Speed Data Samples 250 US 10/US 169 200 I-494 HWY 13 US 169 150 MN 65 US 10/US 61 HWY 14 100 HWY 55 MN 7 US 61 50 0 12 AM 2 AM 4 AM 6 AM 8 AM 10 AM 12 PM 2 PM 4 PM 6 PM 8 PM 10 PM Time of Average Weekday Figure 18. Time-of-Day Trends in 15-Minute Data Samples, By Day of Week Total Number of 15-Minute Samples in 26 Months Averaged across all corridors for each day of the week 100 Number of Speed Data Samples 90 80 70 60 50 40 30 20 10 0 12 AM 2 AM 4 AM 6 AM 8 AM 10 AM 12 PM 2 PM 4 PM 6 PM 8 PM 10 PM Time of Average Weekday Sunday Monday Tuesday Wednesday 18 Thursday Friday Saturday Figure 19. January 2006 Time-of-Day Trends in 60-Minute Data Samples, By Route Total Number of 60-Minute Samples in January 2006 Averaged across all links (both directions) on each corridor 16 Number of Speed Data Samples 14 12 US 10/US 169 10 HWY 13 US 169 8 MN 65 I-494 HWY 14 6 HWY 55 MN 7 4 US 61 2 0 12 AM 2 AM 4 AM 6 AM 8 AM 10 AM 12 PM 2 PM 4 PM 6 PM 8 PM 10 PM Time of Average Weekday Figure 20. January 2008 Time-of-Day Trends in 60-Minute Data Samples, By Route Total Number of 60-Minute Samples in January 2008 Averaged across all links (both directions) for each day of the week 16 Number of Speed Data Samples 14 12 US 10/US 169 HWY 13 10 I-494 US 10/US 61 8 MN 65 HWY 14 6 MN 7 HWY 55 4 US 169 US 61 2 0 12 AM 2 AM 4 AM 6 AM 8 AM 10 AM 12 PM 2 PM 4 PM 6 PM 8 PM 10 PM Time of Average Weekday 19 Figure 21. January 2006 Time-of-Day Trends in 60-Minute Data Samples, By Day of Week Total Number of 60-Minute Samples in January 2006 Averaged across all links (both directions) on each corridor 16 Number of Speed Data Samples 14 12 10 8 6 4 2 0 12 AM 2 AM 4 AM 6 AM 8 AM 10 AM 12 PM 2 PM 4 PM 6 PM 8 PM 10 PM Time of Average Weekday Sunday Monday Tuesday Wednesday Thursday Friday Saturday Figure 22. January 2008 Time-of-Day Trends in 60-Minute Data Samples, By Day of Week Total Number of 60-Minute Samples in January 2008 Averaged across all links (both directions) for each day of the week 7 Number of Speed Data Samples 6 5 4 3 2 1 0 12 AM 2 AM 4 AM 6 AM 8 AM 10 AM 12 PM 2 PM 4 PM 6 PM 8 PM 10 PM Time of Average Weekday Sunday Monday Tuesday Wednesday 20 Thursday Friday Saturday APPENDIX The appendix contains time-of-day speed profiles for the following conditions: 15-minute average speeds – Average weekday speeds from January 2006 through February 2008, all months combined. 60-minute average speeds – Average weekday speeds from January 2007 through December 2007, each month shown as separate trend line. 21