Machine Learning Times
Machine Learning Times
EXCLUSIVE HIGHLIGHTS
Survey: Machine Learning Projects Still Routinely Fail to Deploy
 Originally published in KDnuggets. Eric Siegel highlights the chronic...
Three Best Practices for Unilever’s Global Analytics Initiatives
    This article from Morgan Vawter, Global Vice...
Getting Machine Learning Projects from Idea to Execution
 Originally published in Harvard Business Review Machine learning might...
Eric Siegel on Bloomberg Businessweek
  Listen to Eric Siegel, former Columbia University Professor,...
SHARE THIS:

This excerpt is from Forbes. To view the whole article click here

9 years ago
Ashley Madison And Predictive Analytics

 

We’re all aware of what happened with the “data” stored at Ashley Madison, data that was supposed to be private and protected. This of course is not the first – nor will it be the last – data breach. In fact, it’s safe to say that this will happen over and over again. Put another way, there’s no guarantee that your data – or my data – will be remain protected and private. In fact, guarantees cannot exist in the digital world into which we’re all running.
Apply to AttendDoes this mean that we should all use burner phones, fake email and gift cards we buy (with cash) at Walmart? Does it mean that we should all just assume that we’re victims-in-waiting, that it’s only a matter of time before we’re hacked, breached and shamed?

The Ashley Madison public data dump got everyone’s attention, much more so than the breaches at the OPM, the Pentagon, Target or Anthem. As I write this post, there are families and relationships at risk because for whatever reasons the data is now public. But that’s only one demonstration of risk. Careers, fortunes and health are also at risk. National security is at risk.

What else?

What’s the cumulative risk of what we do on the Web – beyond the obvious hassles around stolen credit cards and the breaches of employee and social security data? Have you ever thought about data breaches and predictive analytics as evil twins? For example, how far are we from a spouse receiving a text that, among other “alerts,” says: “at this very moment, your partner is at the Unfaithful Hotel”? Imagine that text popping up as your changing your kid’s diaper. Imagine that the Unfaithful Hotel is five yards away when the text appears.

Let’s look at what lots of us do on the Web.

  • We search.
  • We Uber.
  • We access healthcare portals.
  • We email.
  • We Facebook and Instagram.
  • We’re LinkedIn.
  • We blog.
  • We tweet.
  • We follow.
  • We shop.
  • We stream music.
  • We invest.
  • We set thermostats.
  • We look for jobs.

What if a data scientist had access to all of the data we create every day, to everything we do on the Web?How could we be blackmailed, cheated and shamed?

Here’s an example.

Whenever I have a medical test I access a healthcare portal and retrieve the test results. I then go immediately to the Web to understand the medical jargon around what cholesterol levels actually mean. I then email some physician friends to share my test results and ask them what they think. They email me back with their interpretation of my test results, with suggestions about, for example, how I need to change my diet. I then order the foods they suggest which are then delivered to my home. I also order some exercise equipment and some natural supplements. I search for some videos that describe what I need to do. If a medication is suggested by my Web search or by my physician, I immediately search for any and all documentation about the medication, especially the side effects and lawsuits that the drug may trigger. If I give the drug my OK, my physician then orders the drug through my prescription provider’s portal to which he – and I – have access. The provider emails a confirmation and requests my credit card number to pay for the portion of the cost that is my deductible. It stores that number for future payments. If the drug is a daily dosage, the provider automatically bills my credit card every 90 days. My credit card notifies me of the automatic charge. The drugs are mailed to me. I then continuously track social media to see what “patients like me” are saying about the medication, its side effects, cost and generic options. Next year’s annual physical starts the whole process all over again, which is now layered on all of the old tests and my entire medical history.

All of this is stored somewhere, everywhere.

How many points of danger are there in this supply chain? If someone nasty had access to all of that information, how could they hurt me?

This is only one supply chain, one communications thread and one customer journey. There are thousands in my – and your – digital history, and there are thousands in our futures. Is there fraud, deception, misinformation or blackmail in your future? Or models that predict business trips, investments, vacations and spending sprees – as well as data about home security systems, especially as more and more of them are Web-enabled?

Might that data be sold to thieves?

Enter the evil data scientist/inferential statistics/data-sharing triumvirate. What might be done with all that data? First, all of the data around digital transactions can be hacked and auctioned on the dark Web to the highest bidder for Bitcoin. That’s a bad day. But on a really mean day, data scientists can construct a profile that can be used in all sorts of nefarious ways. Employers can receive health information while considering employee promotions, information that’s leaked by someone who wants the promotion. Medications can be combined with unhealthy supplements for no good purpose. Incriminating medications can be purchased in your name and “mistakenly” sent to your children, spouses, employers, partners and parents. False digital images can be correlated with actual location data: “was that you in the Uber car leaving the _________ ?”

Predictive analytics is all about the integration and modeling of structured and unstructured data to explain, describe and predict behavior. The more data, the more robust the models – and the Web is the greatest data factory in history.

It’s already happening.

This excerpt is from Forbes. To view the whole article click here.

By: Steve Andriole, Contributor, Forbes
Originally published at http://searchbusinessanalytics.techtarget.com

Leave a Reply