For Our Security, Does the FBI Need a Predictive Model?
Predictive modeling is a process that uses data mining and probability to forecast outcomes. The model is made up of a number of variables about your customers: demographic variables (gender, age, household income, etc.), lifestyle variables (smoker, frequent flyer, etc.) and behavioral variables (last date of purchase, purchase amount, SKU, etc.). Each variable is weighted as to its likelihood to predict a specific outcome (like a future purchase) and a statistical model is then formulated. The model is then overlaid on your customer file and every customer is ranked based on their likelihood to respond to an offer, take your desired action, and even predict the average purchase amount.
If you’ve ever worked with a predictive model, you know it is not static, but an iterative effort that requires constant testing, tweaking and feeding of additional data points. It’s a living, breathing tool that is extremely useful in helping to determine where you should best spend your marketing investment for the highest return.
This same premise could be used to predict the likelihood of terrorist activity — and therefore be a useful tool in our global war on terror.
Think about it for just a minute.
The recent bombing in Manchester, U.K. might have been prevented if only the suspect had been higher on the terrorism watch list.
While authorities noted that the suspect (and his family) were on the list, it was added that there are “thousands” on the watch list and there isn’t enough manpower to track them all. Fair enough. But let’s consider those variables that may have predicted that something was about to happen and that, perhaps, he should have moved higher up on that list.
- The suspects father was linked to a well-known militant Islamist group in Libya
- His two brothers have been separately arrested on suspicion of terrorism offences
- He was reported to authorities two years ago “because he [was] thought to be involved in extremism and terrorism”
- Two friends separately called the police counter-terrorism hotline five years ago and again last year
- Neighbors had called authorities within the last year, noting that the family had flown a flag for a short time that was black and had writing on it similar to jihadists
The final variable is that the suspect had traveled to both Syria and Libya — the latter only a few weeks before returning to the U.K. and launching his attack. Libya is well known as a terrorist hotbed, so add all the previous variables and the "traveled in May 2017 to Libya" variable would probably catapult this guy to the top of the model.
But why doesn’t such a database exist?
Well, privacy concerns, for one. While consumers — and in particular, Americans — argue about their privacy rights, they are already part of every large consumer database whether they realize it or not. If you’ve ever purchased a home, opened a credit card, paid a tax bill, enrolled in a public school, joined a Frequent Flyer program, registered a purchase for warranty coverage, made a political contribution or subscribed to a magazine, you’re in the Experian or Equifax master file.
In many countries around the world, these same kinds of consumer databases exist, so imagine if these files were combined, and then appended with data variables from law enforcement databases and ticket sales from airline databases. Add in databases about weapon and ammunition purchases, and surely there are enough predictive variables that would allow an analyst to build a model that would determine a way to help prioritize security watch lists, and aid in keeping our world just a little bit safer?
Privacy advocates get itchy just thinking about it.
And, of course, there are those concerned about how this wealth of information could be abused, or how hackers could infiltrate and release confidential information.
But as I head through another security check at my airline gate, and I hear more news about losing the ability to work on my laptop or read my Kindle while in the air, I have to think there’s got to be a better way than the seemingly randomization of these security measures. And it seems that a predictive model might be the answer — but since it depends on consumer data at its core, the future is uncertain without it.
A blog that challenges B-to-B marketers to learn, share, question, and focus on getting it right—the first time. Carolyn Goodman is President/Creative Director of Goodman Marketing Partners. An award-winning creative director, writer and in-demand speaker, Carolyn has spent her 30-year career helping both B-to-B and B-to-C clients cut through business challenges in order to deliver strategically sound, creatively brilliant marketing solutions that deliver on program objectives. To keep her mind sharp, Carolyn can be found most evenings in the boxing ring, practicing various combinations. You can find her at the Goodman Marketing website, on LinkedIn, or on Twitter @CarolynGoodman.