A Look at Data-Driven Decisions for Fire Depts. – Part 3

Posted on May 9, 2018
Categories: Data & Analytics
Tags: Fire

As a fire department, are you effective? Are you making defensible decisions for things like staffing, operations, budget? If you answered yes, Captain with the Fayetteville (NC) Fire/Emergency Management Dept. Jason Hathcock and Wilmington (NC) Asst. Chief Frank Blackley challenge you to prove it.

In this series we will review key components to ensuring you’re a data-driven organization: 

  1. Getting Started
  2. Setting Expectations
  3. Data Standards
  4. A Different Approach to Data

In previous blogs we addressed the key components of getting started and setting expectations. In part three we will explain how setting data standards can help prepare your organization for the future. 

Historical data: As an accredited agency, Fayetteville looks at distribution concentration (i.e., arrival of first unit and arrival of effective response force (ERF), or the personnel needed to achieve critical tasks at, for example, a moderate-risk fire). Looking at data historically gives the repartment the information its chief officers need to know what equipment to respond with and what people’s roles should be, for every incident nature code.

Data audits: You’ve heard the expression “Garbage In, Garbage Out.” To reduce the risk of unreliable results, Hathcock strongly recommends auditing data quality. “You can have this whole repository of information there, but if it’s not quality data, you may have a big problem,” one that can’t be fixed overnight, he said. One common problem area: Mixing dispatch and actual findings. “Some departments look at what they were dispatched to; some look at what they found on arrival,” he said. “That matters when you look at data.”

Predictive modeling: Predictive modeling provides insight into when peak demand and lowest demand will occur, not only by times of day but days of the week. Digging deeper into predictive modeling even allows a department to know when a building fire is more likely to occur (and possibly the reason why). “If you had this data, do you think you could justify putting additional units on in a specific time frame?”

Trends: Where do you see increases and decreases in call volume? Hathcock recommends using GIS data: How do they look overlaid on a map? Why? (In Fayetteville, some of the biggest influencers of changes in call volume over time were nursing homes, colleges, and jails.) If call volume has decreased, is it related to a community paramedic program?

Comparing projections to actuals: Hathcock recommends doing a reality check on any projection, especially for something like response times. “I always tell people to look at actual data from responses, because a 4-minute response in a medic unit or fire truck isn’t the same as a 4-minute trip in Google Maps,” he said.

You can go back and read up on part one, Getting Started and part two, Setting Expectations.