Database Reference
In-Depth Information
Gutierrez:
You mentioned that data science was a learnable skill that people
could develop into. How did you do it?
Smith:
Doing the work, conferences, learning from people on Twitter and
other social networks, as well as reading online tutorials. I would say the
best way I learn is by actually doing things. I get inspired to do something and
then I do it. I see what happens and then either iterate on it or do something
else. I think it all starts with the inspiration stuff, which is what I get from
meetings, from sitting in, hearing people and their problems, and kind of
understanding, “Oh, what could I do to fix that? What kind of sources could
I combine? How can we fix this?”
In regards to conferences, I went to maybe one or two for physics in grad
school and didn't really enjoy them. Everyone was doing something so dif-
ferent that it really couldn't be applied to any other project. In data science,
the opposite is true. You can go to a conference and get excited by what
everyone's doing because you can almost always apply what you learn to your
own work. Whether it's a technique that you wonder if you can use with your
data set or an idea of something else that allows you to collaborate with that
person. Going to sessions at conferences almost always helps me improve
my work or the work of someone else, which is great because it leads to an
incredibly collaborative environment. It means that you're not so in depth into
one physical problem that you can't really get out of that hole. Right now, the
field is still at a phase where almost everyone is still interdisciplinary. We're
still well-rounded and can take applications from anywhere and still have them
work. So conferences are a great treat.
The only downside to a conference is that you come back with a mile-long
list of things to try, and you eventually realize that you can't try every single
thing. You get a high from the people doing great things that are applicable to
your work, so it's tough to come back and have to prioritize what to try out.
Also, when you are at a conference, you aren't working, so there's always that
to think about before deciding to attend as many conferences as you would
like to attend.
In regards to Twitter, I am constantly scouting for new people to follow to add
to the collection of the quality people I already follow. I love to read ideas and
ask myself if they have insights that I can use in my own work. After that, it's
Google and Stack Overflow. I do this because it's very different from reading
a book. Most of the topics I'm used to reading are textbooks, which lead to
the thought pattern of, “Oh, okay, yeah. I totally get that.” And then you're like,
“Wait. How does my data relate to these abstractions and those equations?”
And so I think it's helpful when you find tutorials on the Internet that can
show you, “Okay, this is my data set. This is how I relate it to these equations,
and this is how it looks in code.”
