More Confidence in Your Traffic Data

Today’s press release from Traffic.com has one thing I’ve not seen before in our industry: geospatial information with a confidence rating. This being traffic data, the confidence information relates to the “accuracy” of the predictions.

The Traffic.com Predictive Engine applies statistical pattern-matching algorithms to current real-time conditions and historical travel models to estimate future speeds and drive times. Based upon the characteristics of each data sample used to calculate drive times, a ‘Confidence Factor’ is derived providing a percentage level of accuracy for each travel time value. With this sophisticated model, application developers will benefit from unparalleled control of end-user content in consideration of extended geographic coverage.

Getting more data, or in this case metadata, is generally thought of as a good thing. The exciting part is how Traffi.com’s customers will (1) monetize it (via advertising, subscriptions or a mix) and (2) present it to end users. All three business models are in use (NAVTEQ, which now owns Traffic.com offers subscriptions only; Microsoft uses ads and subscriptions, for example) and it remains to be seen which if any, are the long term winners. Of more interest to me is how the new confidence information will be presented the end user. Will it be predigested so the end user sees only the end results of how company algorithms munch the data? Will end users have some say in how the confidences are used? I’d suggest that will be parallel to how in some routing apps end users can request to “stay off the highway”. Will some providers offer complex interfaces to let end users manipulate the data in a very granular fashion? The folks at Traffic.com are curious, too, based on a conversation with them last week.

The other interesting point to consider: algorithms to use the confidence information may well need to change from geography to geography. Experience may show that urban data (of which there is more historical information) has more or less accurate confidences, than say rural data.

Finally, if you ever wondered: Your current speed is not the best predictor of your future speed on a route. Density and volume are! Happy travels.