The science of predicting traffic

Predicting traffic has always been a tricky proposition. Most “real-time” traffic web sites tend to have some lag time between retrieving data from sensors embedded in roadways and updating web sites showing traffic conditions.

The above video explains a system being tested in the Bay Area that predicts what traffic will be like in the coming hour. People who sign up can receive an email or text message before they leave home or work telling them what traffic they are likely to encounter on their regular routes.


The test is a combined effort by IBM, Caltrans and UC Berkeley — with IBM calling this a patent-pending technology on their website. The system is also explained at Welcome to the Fast Lane, the blog by U.S. Transportation Secretary Ray LaHood.

The big question: how useful is such a system to the average commuter?

Let’s say Joe Commuter received a text that their usual freeway route was more congested than usual. Seems to me the system is most beneficial if Joe Commuter has a good alternate route, the alternate route doesn’t get too congested with others seeking an escape route from traffic and/or Joe Commuter has the flexibility perhaps to alter their commuting time or take transit instead.

The other thing worth noting here is that IBM is obviously very interested these days in the field of transportation management — check out their main web page about the company’s efforts. IBM is American-based and I think given this country’s transportation challenges (too numerous to list or repeat here), it’s intriguing to see an American firm trying to compete globally in the transpo field.