Database Reference
In-Depth Information
And often actually, one of the dirty secrets is that a lot of the time it's not
just that there's an algorithm that solves the problem. There's also a human
workflow of some kind involved. You can actually find this out if you start
picking at different big algorithms that people have and use. You find that
there's often a stage where you have analysts checking things or somebody
is sending off a data set to a Mechanical Turk. That means there are a lot of
options for how you might go about it. Maybe you have some people who
could do a bit of the work, and then you have the algorithms and the data, and
then you have the problems. So piecing those together and imagining how
they'll come together—that to me is where you get the magic.
But then after that, within each one of those pieces, there's lots of hard
work, and often you get really interesting stuff happening. So there are some
fascinating algorithms to play around with and really insightful things that
people do when they have brainy insights where they say, “If I did it exactly this
way, something cool would come out.”
Gutierrez:
What advice would give to someone starting out and what should
they strive to understand deeply?
Heineike:
I think perhaps they would need to start by looking at themselves
and figuring out what it is they really care about. What is it they want to do?
Right now, data science is a bit of a hot topic, and so I think there are a lot
of people who think that if they can have the “data science” label, then magic,
happiness, and money will come to them. So I really suggest figuring out what
bits of data science you actually care about. That is the first question you
should ask yourself. And then you want to figure out how to get good at that.
You also want to start thinking about what kinds of jobs are out there that
really play to what you are interested in.
One strategy is to go really deep into one part of what you need to know. We
have people on our team who have done PhDs in natural language processing
or who got PhDs in physics, where they've used a lot of different analytical
methods. So you can go really deep into an area and then find people for
whom that kind of problem is important or similar problems that you can use
the same kind of thinking to solve. So that's one approach.
Another approach is to just try stuff out. There are a lot of data sets out there.
If you're in one job and you're trying to change jobs, try to think whether
there's data you could use in your current role that you could go and get and
crunch in interesting ways. Find an excuse to get to try something out and see
if that's really what you want to do. Or just from home there's open data you
can pull. Just poke around and see what you can find and then start playing
with that. I think that's a great way to start. There are a lot of different roles
that are going under the name “data science” right now, and there are also a
lot of roles that are probably what you would think of data science but don't
have a label yet because people aren't necessarily using it. Think about what
it is that you really want.
