Menu
Menu
Fire and Rescue NSW to tap into predictive data models, machine learning

Fire and Rescue NSW to tap into predictive data models, machine learning

The emergency services organisation wants to use predictive analytics to pre-emptively deploy resources

Fire and Rescue NSW is tapping into predictive data models and machine learning to better prepare its resources before emergency situations occur.

Speaking at a CeBIT conference in Sydney, the CIO of the safety organisation, Richard Host, said the response speed of emergency services can mean life or death for people. In order to improve the speed with which the organisation can deploy resources, Fire and Rescue NSW has been working on a system dubbed ‘Miinder’.

Miinder assesses every single property in the state in conjunction with the weather forecast and predictive data feeds to develop a model that shows where resources are likely to be needed in order to deal with an emergency, he said.

“We all know intuitively that the last day on school holidays there’s going to be an accident on the M1. We know intuitively that some kid in the school holidays is going to light a fire. But what if we could look at this systematically and in a far more intelligent way?" Host said.

The project is currently in a proof of concept phase. The first iteration of Miinder will run complex algorithms over aerial photography to measure how close different types of vegetation are to houses and assess the risk of fire.

The CIO said he will look at integrating weather data into the heat map and adding features that allow a user to replay past incidents and model what might happen in the future.

Host will look at integrating other risks besides bushfires, and develop a mobile app that can score a house’s chance of suffering an incident out of 100. Users would be able to look at the score for entire streets, suburbs or cities.

Based on Miinder’s calculations on where incidents are most likely to occur, Fire and Rescue crews could be pre-emptively sent to specific locations during certain times of the day.

“By placing a crew there means the proximity to the crew brings the risk down; sending resources out in advance will reduce the overall risk,” Host says.

“And Miinder will go back and monitor; it will go back and say… ‘I really should have sent this crew, not this crew. If I sent that crew, I would have had the optimal outcome. I’m now changing my rule and I’m now going to use that in the future. So I’m going to start learning’.

“Also, it will, for the first time, show a real connection between the spend on resources and the consequence on safety for the community. For every extra million spent there, you will see through the formulas the outcome. So we can stop over engineering here and under engineering there.”

Fire and Rescue recently purchased SAP’s HANA in-memory database to support Miinder's data crunching. “If you use it properly it will run up to 1000 times faster than a traditional database,” Host said.

Join the CIO Australia group on LinkedIn. The group is open to CIOs, IT Directors, COOs, CTOs and senior IT managers.

Join the CIO newsletter!

Error: Please check your email address.

Tags predictive analyticsCeBIT 2014Fire and Rescue NSWbig datamachine learning

More about CeBITSAP Australia

Show Comments
Computerworld
ARN
Techworld
CMO