When we discuss the topic of traffic accidents, most of the time the "accident" part of the discussion is due the vehicle crashes occurring in unpredictable locations and to random motorists. Of course, in the strictest sense, they are not random. If they were, insurance companies would no ideas on how to price their products. They use mathematical models to determine risk.
In the same way, systems are being developed that permit law enforcement to predict locations with the greatest likelihood of car accidents on a given day. One program being tested in Tennessee is called Crash Reduction Analyzing Statistical History (C.R.A.S.H) and uses data calculate traffic risks within the state.
The program has been in place for six months and has achieved an accuracy rate of 72 percent. It uses five-by-six-mile squares and examines the data for that section in four-hour blocks, issuing a probability of crashes within that sector, something like a weather forecast.
It is a type of predictive analytic technology, which is similar to systems that other law enforcement departments use to predict criminal activity. This is one of the first programs to use it for traffic control.
The program can sift through vast quantities of data, discarding information that is irrelevant. It has also been used to identify locations where intoxicated or drugged drivers may be present.
It is not perfect, but then neither are veteran state troopers when it comes to allocating where law enforcement resources will be deployed. It gives the state patrol one more tool to better enable it to try to have troopers in locations where they will be needed.
Ultimately, the real goal of the program is to save lives, and in the six months it has been in use, traffic fatalities have declined 5.5 percent from the same time last year.
The program should improve with use, as more data is available and the program can be refined to recognize which information is most useful.
Source: Government Technology, "Software Predicts When, Where Accidents Occur on Tennessee Highways," Shelly Bradbury, McClatchy News Service, August 4, 2014