By Tom Still

MADISON – Data science is a fancy term for statistics. It’s
the extraction of knowledge from data, which can be derived from multiple
digital sources and turned into a resource to make business decisions.

Sounds simple enough, right? But the complex formulas behind
data science can involve artificial intelligence, information theory, machine
learning, computer programming, data warehousing and high-performance
computing, along with other disciplines most people hadn’t heard of a decade
ago.

Another way of thinking of data science is algorithmic math.
It’s about using a precise sequence of operations to produce a finite outcome…
a concept that sets many geek hearts a flutter, and which increasingly drives
consumer and marketing decisions affecting everyday life.

It’s also a trend that raises questions about where data
science ends … and where human choice and analysis begins.

Automated decision-making is meant to take the human out of
the equation. It’s less emotional, more statistically efficient and
increasingly more predictive of human behavior, especially when the right data
– and enough of it – is analyzed.

The potential applications are seemingly endless, from loan
approvals to weighing insurance risks, and from medical diagnosis and public
health to crime prevention and sentencing.  Data science can be used to
determine how to market products to targeted audiences, to set insurance
premiums based on the health habits of the insured, and to predict how
financial markets will perform.

Skip the gym today? Your insurance company might someday
know that.

Data science is behind Wisconsin-based startups that are
changing how many people shop for groceries, to cite one recently reported
example. Fetch Rewards, GrocerKey Inc. and Pinpoint Software Inc. are bringing
digital technologies to the grocery industry. Technologies developed by these
Madison-based companies can be used to deliver groceries, compare prices, order
food, track inventory and expiration dates, and connect customers with
point-of-sale coupons. 

In the process, these platforms create data banks that are
potentially valuable to companies that want to know more about your behavior
when you’re shopping. At the end of the day, that’s the core value behind data
science companies… the data.

If it’s groceries, no one really cares if you buy organic
oat bran or jelly doughnuts – unless the data is sold to your health insurance
company, which could chart how well you’re following that low-sugar,
low-cholesterol diet.

But what if data science is used to determine whether you
qualify for a loan? By using social media connections, or even examining how
people fill out online applications, big-data-driven lenders can know borrowers
at a deeper level.

A recent New York Times story examined how data science
might be used to fine-tune assessments of whether a loan applicant is a solid
risk or a potential deadbeat. That can help people who deserve the benefit of
the doubt, but hurt those who come up short on a data-driven analysis that was
lacking human review.

“A decision is made about you, and you have no idea why it
was done,” Rajeev Date, an investor in data-science lenders and a former deputy
director of the Consumer Financial Protection Bureau, told the Times. “That is
disquieting.”

Data science is still a young science, subject to human
error – or, more precisely, human judgment about what data is crunched in the
process. The promise of data science, however, is that software algorithms can
learn as they go by sifting through ever-increasing amounts of information.

Because that information originates with humans, is analyzed
by humans and applied by humans, data science may never lose its human touch.
It’s less a question about the data itself than how it’s used.

As the world of Big Data grows, so will opportunities to
address truly global issues such as climate change – as well as routine
consumer choices in the grocery aisle. As with any science, however, data
science will carry an obligation to get it right and present all results,
successful or not. The data itself may be complicated, but transparency in how
it is presented will allow people to make their own decisions.