Predictive technology helps speed roadside assistance CAA uses analytics to forecast where road issues are likely and to reduce wait times
Canadian Automobile Association members around the GTA might be cursing their car for a significantly shorter time if it won’t start, gets stuck or otherwise becomes immobilized this winter.
That’s because of a new predictive technology the association’s South Central Ontario branch has deployed that can reduce wait times by 15 to 20 per cent by moving trucks to locations they’re more likely to be needed.
“It’s an advanced predictive analytics engine that takes data in real time ... to help deploy our trucks better,” says Tony Tsai, the CAA’s assistant vice-president of communications.
“It takes things like traffic patterns, weather, as well as our trucks’ GPS information and combines it with our historic data to be able to predict the severity, frequency and types of breakdowns that can occur in a particular geographic area,” Tsai says.
Known as Geo-Temporal Gen 2.0, the system has been up since the summer and has enabled the CAA to cut an average of 11 minutes off their 30-to-45-minute response times, he says.
But the real test of the technology — which the association pioneered — will come as winter descends in earnest, bringing battery-sucking cold, ditch-dealing ice and car-trapping snow drifts.
The system — which employs brute computational power and learning technology — drills deeply into the day’s conditions, taking things as specific as dew points, wind speeds and directions and barometric pressures into account.
“We factor that along with road conditions, traffic patterns, road construction and our own trucks and their rate of travel,” Tsai says.
It will then show dispatchers at the CAA’s Thornhill depot where in the coverage area their trucks will likely be needed most.
About 90 per cent of those calls will be for minor problems, such as lockouts, dead batteries and flat tires, Tsai says.
“This technology allows us to kind of be able to see those types of breakdowns and identify if there’s more likelihood for battery breakdowns or lockouts,” Tsai says.
“It also predicts the likelihood of breakdowns occurring in hard to reach places ... like underground parking lots. It can get that specific.”
Tsai says the association’s south central branch has 1,200 trucks in its rescue fleet and that these can be deployed in greater concentrations to the areas the system says they’ll most likely be needed.
The system is hindered, however, by municipal licensing regulations that limit the number of trucks the association can deploy in some of its coverage areas.
“For example, Mississauga has very strict licensing so we can only ... use so many trucks that are licensed in Mississauga to deliver service (there),” Tsai says.
So far the system has a 98 per cent accuracy rate in its ability to predict areas that will need more service.
The CAA’s South Central Ontario branch services an area stretching from Kingston to Sarnia as well as Sault Ste. Marie.