How Two Real-Time Analytics Techniques Help Public Safety Agencies Predict the Future

Intergraph InSight Suite

In a recent post, I discussed how many public safety agencies are becoming data-driven decision makers. And it’s true. Over the last five to 10 years, advancements in business intelligence like real-time data streams for analytics and machine learning are helping public safety agencies better predict the future based on historical data. This in turn, makes communities safer and resource management more efficient for agencies.

Continuing these trends and topics, I spoke with Hexagon Safety & Infrastructure’s Chris Klimm, principal consultant for our Intergraph Business Intelligence for Public Safety solution. During our conversation, Klimm shared several examples of how public safety agencies could immediately apply real-time data streams to analytics and machine learning. Here’s what he had to say.

Stream Analytics
Stream analytics takes data from external sources and pushes it to a dashboard in real time. This type of reporting is very useful for users that oversee multiple, active resources at one time, like supervisors. Here are a few ways agencies can use a real-time data stream for analytics:

  • Real-time data tracking for viewing where all units are at any point in time
  • Real-time information to identify health outbreaks and epidemics by analyzing locations and onsite symptoms of patients
  • Real-time information to identify major attacks and terrorism by analyzing social media and recognizing patterns associated with incoming events, which can be quicker than a human

Machine Learning
Machine learning is a type of analysis that uses historical data to predict future outcomes. It enables computers to learn from data and experiences and to act without being explicitly programmed. Analytic models, like the ones found within Azure Machine Learning Studio, are trained using real historical data. Then, the models are evaluated by holding some of the historic data back to test against real data.

The example Klimm used within a public safety context was false versus true burglar alarms. More than 90 percent of the alarms sent to police departments are false alarms. Since every alarm needs to be responded to and more than 90 percent of alarms are false, it can ultimately be determined as an inefficient use of limited resources. However, machine learning can be used to predict if the alarm is a true or false alert. It’s not a perfect science, but it can help dispatchers prioritize which events they send resources to. When testing machine learning, it’s important to test different models and variables to find a solution that fits a given scenario. One of the most significant things to consider is how vital clean data is to the machine learning process.

If you would like to learn more about real- time analytics and see examples of how it’s being used around the world, check out our HxGN LIVE conference! There will be training opportunities and sessions that focus on business intelligence for public safety agencies as well as solutions featured in The Zone.


Pamela Van Asseldonk
About Pamela Bernier
Tags: