BY JESSE ROMAN
IF A BRIGHT RED BEACON would light the top of every house without a working smoke alarm, fire safety outreach and education campaigns would be a whole lot easier. However, the big questions such as “which areas of the city should we target?” and “what messaging works the best?” is still a guessing game for most fire departments. For others, big data is starting to yield interesting answers.
, Las Vegas, June 13-16, 2016
Predictive Community Risk Reduction -- Using Data Science to Reduce Fires
Wednesday, June 15, 8:00 AM to 9:00 AM
Matt Hinds-Aldrich, Atlanta Fire Rescue Department; Bistra Dilkina, Georgia Tech; Mike Flowers, Enigma.io; Bill Schneider, Experian Marketing Services; Nathanial Lin, NFPA
Last year, fire departments in Atlanta, Philadelphia, and Los Angeles got a boost in this area from the marketing firm Experian, which used its massive vault of consumer data to map out where fire safety outreach is likely to have the most impact.
Experian maintains huge quantities of consumer data on households and individuals—everything from age and income, to education level, marital status, how long someone has lived in a residence, if they own or rent, and much, much more. All of that information is run through sophisticated computer algorithms to make inferences about a person’s tastes, and habits. Advertisers use Experian’s data to trailer their marketing campaigns to target specific segments of the population. This is how they know to place a billboard in a certain street, for instance, or send mailers to particular zip codes, or advertise on certain radio stations. The marketing data even informs messaging: how to say something to get the most impact. Data scientists at Experian thought that all of that could also be used for public service campaigns.
So the company obtained two years of National Fire Incident Reporting System (NFIRS) data in the three cities, to see the locations of actual fires, explained Bill Schneider, a solutions support consultant at Experian who worked on the fire project. “We are able to take the addresses where fires occurred and match those to our consumer profiles to identify the characteristics of people living at that address,” he explained. “Like we do in the marketing world, we then developed a statistical model to compare the characteristics of people living where a fire occurred to the general population. In this case we treat the household where fire occurred like a consumer for a marketer, and apply the same types of statistical analysis to determine the most differentiating characteristics in this population.”
By using this methodology, Experian’s data scientists identified all of the statistically significant variables and characteristics that seem to make certain households in each city at higher risk for experiencing a fire. It ran all of the data it had for each household through the models, and plotted the results on a map—higher risk areas were colored in red, and lower risk areas in green—allowing fire officials to see exactly which streets in which neighborhoods were likely to benefit most from a public outreach effort. It’s not a beacon on each house, but it’s about as close as you can get.
“Intelligent door knocking is the crux of this thing—firefighters knocking on doors talking to residents about fire safety has been proven to be way more effective than sitting at grocery store talking to random people who pass by,” said Matt Hinds-Aldrich, a member of the Assessment & Planning section of the Atlanta Fire and Rescue Department. “Intuitively I know where the challenging neighborhoods are and where we should focus, but I don’t know which street I should be on, what door to knock on, or what expect. Do they speak Spanish? Are they likely to have children? Are there adults with limited mobility there? How many people live there?”
Experian can now provide that information and even coach firefighers on what types of messages these groups are likely to respond to. For its marketing business, Experian sorts households into 71 different mosaic segment groups. Each mosaic group shares a wide number of characteristics that make those households very similar to each other, but different from other segments of the population. This helps advertisers trailer their messages even more, and the same has been done with the fire project. After Experian identifies which mosaic segments have the highest incidents of fire, “from that we can develop communications for awareness and prevention that are more customized to that type of household,” Schneider said. “If household has a single parent on the lower end of the economic scale, that is very different than a retired senior living on very limited income. You would develop the messaging very differently.”
Experian has validated its fire risk model and the information is promising, but as of March it is all very conceptual. The company is working with fire officials in Atlanta on how it might implement this new knowledge.
“It’s still in the early stages. We’re very excited about the results, now we need to have discussions about what action items can be taken as a result,” Schneider said. “We’re really still in the educational phase of this. The math looks good, it’s a matter of developing programs that best take advantage of the information.”