Can predictive analytics reduce roadway collisions in Indiana?
Connecting state and local government leaders
The state's new Daily Crash Prediction Map calculates the likelihood of collisions occurring in a specific area based on weather, traffic and road conditions.
When the Indiana State Police learned that roadway fatalities in the state increased 7.5 percent from 2014 to 2015, it asked the state’s Management and Performance Hub to help reduce the risk of collisions.
The online Daily Crash Prediction Map is a joint effort to make roads safer for citizens and decrease emergency response times. The map is interactive and public facing, and calculates the likelihood of collisions occurring in a specific area based on weather, traffic, road conditions, the time of day, historical information and census data, according to a video by the state police.
The Management and Performance Hub, which provides custom analytics solutions to state policy questions, worked with KSM Consulting, a provider of technology and data analytics services, to use the historical data to build a machine-learning model. New information is fed into the model every day to generate the predictions on the map, according to MPH Analyst Ashley Schenck.
The average annual daily traffic volume data comes from the Indiana Department of Transportation, and census and county business pattern data is pulled from Census.gov. The state police and law enforcement agencies across the state collect crash data daily on a statewide database and collision-reporting portal, the Automated Reporting Information Exchange System. The map uses this data dating back to 2004, according to 1st Sgt. Sergeant Rob Simpson.
Users can choose any three-hour window for any city or address, and the map will display the areas of risk on the road. The tool uses color shading to show a low, moderate or high probability of a crash occurring in each 1-square-kilometer area in the state. It also shows relevant historical crashes as dots, which represent fatalities, property damage or incidents requiring emergency medical service response.
Zooming in on the dots provides specific information about the accident, including what led to the accident and the cars and injuries involved, according Simpson. Along with improving road safety, the map is intended to reduce emergency response times to accidents by allowing officials to pre-position responders in areas with high probabilities of crashes. Citizens can also use the tool to better plan travel routes, especially around the holiday season.
The first phase of the predictive map launched on Nov. 9, and MPH will continue to evaluate the tool based on user feedback and input. According to Schenck, the use of near real-time or real-time data could be considered for a future phase.
Other enhancements may include making the map into a mobile application. As is, the map is accessible by mobile device browsers. And because the tool is a learning algorithm, 1st Sgt. Simpson said, the map is always receiving data and making adjustments to provide more accurate predictions. “We want to continue to improve and maybe offer a little bit more robust system as we continue to move forward,” he said.
Indiana’s tool is similar to apps developed by Kentucky and the Federal Highway Administration that use connected devices and weather data to improve road safety.