The year is 2054 and crime as we know it has been virtually eliminated.
With the help of 3 mutated humans, called “Precogs”, capable of seeing crimes before they happen, the “PreCrime” unit, led by Tom Cruise, can stop murders before they occur, presumably without messing up their perfect hair.
Now while the hair might be real, obviously the plot of Minority Report is science fiction. While a system, human or otherwise, capable of seeing what has yet to unfold would be a dream, as it stands in 2016 and for the foreseeable future, we possess neither the science nor the technology to predict events with any certainty.
But that doesn’t mean we’re not going to try.
If we think about George Santyana’s famous quote, “Those who do not learn history are doomed to repeat it”, or Confucius saying “Study the past, if you would divine the future”, the message becomes clear.
If we know what’s happened before, we can guess what will come after.
There’s a good chance that if I were to describe to you the actions of an individual, you would be able to tell me what time of day it is. If I tell you that a person has just taken a shower, they’ve brushed their teeth, shaved, gotten dressed, and is currently drinking a cup of coffee, you understand from those behaviors that it’s more than likely morning.
We understand, based on our own past experiences, that when these behaviors are grouped together in a short period of time, it’s right after we wake up and start to get ready for the day. From there we can infer other things if given more information. If this person shaved their face, it’s likely a man. If I told you that he put on a suit, we can infer that he’s going to work.
By studying these generalized patterns in behavior, we can guess with remarkable accuracy what the next likely outcome will be. And as I demonstrated, the more information that we have the more accurate we can be with our predictions.
We live in an age where our interactions with technology generate a massive amount of data and I’m not just referring to texting or social media profiles.
It’s called Big Data, and if we just take a look at your daily routine, you can start to get an idea of just how ‘big’ it is. Using me for example, this morning upon getting to work, I used my badge to get in my building, used it again to get in my office, then I logged into my computer, clocked in, and started working for the day.
In that brief time, I generated gigabytes of data by appearing on security cameras, creating a scan time into the building, a scan time into the office, windows log with the time I logged in, a clock in time on my time card, an internet search history, new save time on my documents, and the list goes on and on.
Predictive Analytics is the science of gathering this data (called “data mining”) from data lakes, NoSql databases, or with Hadoop, teaching a machine what to look for (machine learning) and creating a predictive model to determine the likelihood of a future or otherwise unknown event.
By analyzing past events, we can see patterns that form allowing us to make an educated guess in order to determine future behavior.
But predictive analytics isn’t a new concept, and in fact, has been used for various purposes for quite some time. Credit bureaus, for instance, uses predictive analysis to generate a model of your payment history to determine the likelihood that you’ll make a payment on time in the future, which is expressed by your credit score. Insurance works the same way, where an analysis of your driving record will determine your risk of getting into an accident and how much you’ll have to pay.
What’s new about it is the realization of how much more we can do with it.
Predicting your behavior.
You’ve likely heard the news story about Target, where a young girl purchases diapers, wipes, and baby formula, at which point Target saw her purchases and mailed her coupons and marketing offers that one would deem useful to an expecting mother. She ultimately had to tell her parents that she was pregnant after her father found the coupons and berated the store’s manager for their “assumptions”.
Target, using predictive analysis, saw that she was buying things that are typically needed when having a baby and, in the hopes of making her a return customer, provided her with marketing materials that would help her save money for the things she will likely have to buy in the future.
The media uproar that was created from this incident brought predictive analysis into the public eye and emphasized its power, albeit in a biased and extraordinarily one-sided manner.
The fact of the matter is, just like you wouldn’t get angry at a detective for using the clues left behind by criminals to solve a case, a store shouldn’t be targeted for using the clues left by their consumers to make more money, even if the logo of the store in question is a giant bullseye.
But that aside, if the predictive analysis is good enough to predict that a girl is about to have a baby, what’s to stop it from being that detective and stopping criminals?
The Future of Compliance.
Finance is a highly regulated industry, due to the role its members play in maintaining a stable global economy. Currently, the Financial Industry Regulatory Authority, FINRA, the Securities Exchange Commission, The SEC, and the Commodity Futures Trading Commission, CFTC, specifically require that the transactions and communications of traders, among others employed at a financial firm, be monitored, audited, retained, and archived for a length of time.
This is a good solution for identifying illegal actions like insider trading after they occur. For this scenario, surveillance allows us the ability to see that a person received an email telling them a company is going to announce bankruptcy, on Tuesday at 1PM, and then they proceeded to sell all of their shares shortly thereafter, at 1:30 PM.
A downside of simple surveillance however is that, for example, while we can see an explicit email from a customer complaining and saying that they’re leaving the company, it doesn’t allow us to do much about it. The damage has already been done, and there’s no way anyone outside of that customer relationship could have seen it coming.
However, if we use predictive analysis on just an individual’s communications, by training the machine to look for emails that indicate negative emotion or any deviation from their normal rapport, we could detect that customer relationship degradation over the course of time. A warning like this would allow a supervisor to intervene possibly preventing the client from leaving the firm.
But let’s refer back to that detective stopping crime.
In June of 2013, Bloomberg News reported that something was amiss in the Big Data. Reports of front-running, collusion, and market manipulation surfaced, which involved individuals in several banks conspiring to execute Forex trades that would change the market in a predictable manner, which they would then take advantage of for personal gain.
As I mentioned before, predictive analytics, at its most basic, is using data to teach a machine what is and isn’t normal. Not only could predictive analysis have been used to send up a red flag at any of these firms when people started deviating from their normal trading habits, but a properly trained engine could have also caught the chat rooms where they discussed “the Cartel” and “whacking the market”, as these terms would have deviated from their normal communications.
There is little to no debate that predictive analysis is powerful, as it is useful, but is it the future?
Catching financial criminals before they can cause real damage or even preventing another global financial crisis, should be the top priority for financial companies and regulators alike. And predictive analysis has proven time after time to be an effective solution, limited only by the imaginations of the data scientists behind it.
While it’s not the future where humans, capable of precognitive visions, tell Tom Cruise to arrest Jesse Litvak or Martin Shkreli before they show up for their first day of work, it’s my belief that predictive analytics has the ability to gift us with new insight and perspective into approaching compliance surveillance and risk prevention.
But I’m willing to bet you knew I was going to say that.