Database Reference
In-Depth Information
Intuit
is a key word here. Like the vision of the potter or the vision of the artist, in-
tuition and creative thinking are essential to the Data Science process: a key and uniquely
human attribute without which the process cannot and will not succeed.
Does intuition make for infallible results? No. Per Mark Whitehorn, holder of the
University of Dundee's analytics chair: “We're not in that territory. Of course, a good Data
Scientist should be able to give you a probability - 'we think this is likely to be correct in
95% of the cases, 92% of the cases' and so on.”
Despite imperfections, humans outrank computers in one key area. Computers are
great for handling purely functional questions, but they are lousy at conjecturing and fig-
uring out the “why” of data. They simply cannot do it. At least not yet. And even when
artificial intelligence software is eventually developed to help machines at least begin to
contemplate the “why,” they will still need a great deal of manual human intuitive help. (A
later chapter will look at fledgling artificial intelligence approaches and software heading
us toward a place where at least
some
intuition will be possible on the part of machines.)
A minority of observers (especially those from the traditional scientific community as
opposed to those from the business, social research, or marketing research communities)
are deeply skeptical of Data Science as a discipline. Furthermore, they are skeptical espe-
cially
because
of the human factor.
Matt Asnay, VP of mobile at Adobe, notes that since the role of the Data Scientist is
to impose “the right questions” on data, this very exercise brings with it a bias which can
contaminate the resulting BI. Statistician Nate Silver adds: “[Big Data] is sometimes seen
as a cure-all, as computers were in the 1970s. Chris Anderson … wrote in 2008 that the
sheer volume of data would obviate the need for theory, and even the scientific method. ...
These views are badly mistaken. The numbers have no way of speaking for themselves. We
speak for them. We imbue them with meaning [and] we may construe them in self-serving
ways that are detached from their objective reality."
But this logic belies one key fact.
Not all bias is bad
.
Bias informed by experience and knowledge of the topic area being explored is our
friend. Bias based on practical, pragmatic analysis of real world situations and real world
information needs is our friend. Fact is, there's a subjective aspect to Data Science which
does not exist in fields involving the practice pure-science - fields such as chemistry where
one is looking for strictly defined and provable empirical results.
The
informed
and rational bias of of the Data Scientist in the process of unearthing,
combining, and imposing questions on data is not only a valid aspect of the overall equa-
tion, but a
fundamentally necessary
element of the overall equation. Bottom line: Data
