At Replica, we’re thrilled to share the results of Texas A&M Transportation Institute (TTI)’s recent validation of Replica’s Average Annual Daily Traffic (AADT) and hourly volumes datasets. This evaluation, which also included Inrix's AADT and hourly volumes estimates, inspires confidence in using these datasets for valuable insights into granular traffic patterns.
TTI's validation consisted of a detailed comparison of AADT and hourly volumes estimates from both Replica and Inrix against observed data collected by TxDOT’s statewide permanent count stations.
The findings were very encouraging, as both Replica datasets had strong correlation to observed counts. Replica’s AADT dataset resulted in a Mean Absolute Percentage Error (MAPE) of 12.1%. Similarly, the more granular hourly volume estimates showed a strong correlation, with a MAPE of 23.7%.
TTI’s findings underscore the close alignment between Replica's AADT and hourly volume estimates, and TxDOT’s observed data. This level of accuracy is crucial, as State Departments of Transportation need data they can trust for all road segments, and collecting this data for the whole network manually is especially challenging and cost prohibitive.
Many DOTs rely on a combination of permanent counters, short-term counts, and default values for various segments, particularly for lower-volume roads (functional classes 6 and 7), to address these challenges. However, these default values, often derived from median counts on similar roads, have been found to have significant errors.
Replica’s dataset offers a cost and resource-effective alternative, and TTI’s work validates this approach.
The thorough and insightful assessment conducted by TTI not only validates our datasets, but also supports Replica's commitment to iterative innovation, advancement, and improvement. We are firmly committed to consistently refining and validating our models as part of our broader mission to instill trust in Replica data and expand its applicability to more and more use cases.
Want to learn more about Replica’s AADT data? Reach out to us here.
