Big data is everywhere, it seems. From privacy concerns raised by the NSA’s collection of phone information, to hackers stealing more than a billion passwords, the accumulation and processing of ever-greater amounts of data are shaping our world.
We think nothing of it when Amazon suggests we may also like a similar product or Facebook suggests other items for us to “like.” Some use of this type processing data in a predictive model may help create safer roads. The Tennessee Highway Patrol (THP) is one of the first in the nation to use data to drive, literally, where they deploy their officers.
A new software program, known by the acronym CRASH (Crash Reduction Analyzing Statistical History) is used to process a wide variety of data, like traffic patterns, football games, weather and other factors that affect traffic safety.
The system then calculates probabilities for geographic areas that may see increased car accidents or other adverse events. It can be used to target areas prone to drug and alcohol use that could trigger an increased likelihood of DUIs.
It uses information like locations with Alcoholic Beverage Commission licenses, large sporting or music festivals, which possess the potential for more incidents of drunk driving.
However, it is not infallible. During this pilot phase by the THP, it has been accurate about 72 percent of the time. And while it suggests geographical areas where there is a greater probability of potential crashes, it has its limits.
Individual drivers within a projected area would still have been identified with by the troopers, and their driving conduct would have to provide the trooper with reasonable suspicion before a traffic stop would be authorized.
Source: Chattanooga Times Free Press, “Software Predicts When, Where Accidents Occur on Tennessee Highways,” Shelly Bradbury, August 1, 2014