Database Reference
In-Depth Information
So now a company with data scientists wants to hire a data scientist. The data
scientist gives the company one of their projects and now the company is sup-
posed to judge how good the data scientist's model is? The data scientist does
not even know it. The company does not know what the data scientist really
did or could have done. And, unless it is in the same industry, the company
may not even really know the data either. So, looking at a project from the
semi-outside, even from a managerial perspective or hiring perspective, and
saying how good it really is in terms of what it could be or should be is almost
impossible. And that directly translates into also measuring the skill set of the
data scientist because that is just how good a project can be produced. So
quality control is a problem of the project and it translates into a real difficulty
on evaluating the skill and the proposal of somebody you are hiring or talking
to. It is easier if you already have a crew that can look at it and have a smell
test on the data scientist as well.
Gutierrez:
What do you think of the present and future of data science?
Perlich:
In the present, I am not quite sure where we stand as a group. Part
of this uncertainty is that the narrative of data science and big data has really
been shaped a lot by people who have a stake in the game, whether it is ven-
dors or consultants. They focus on telling you all the cool things you can do
with big data. This has generated expectations that a lot of people struggle
with fulfilling, which is a negative.
At the same time, I think it is very good to have much more awareness of the
opportunities that come with data. I think that is a great thing, because we
have been arguing about data and the future of it for the last twenty years.
I am glad somebody is finally listening. I think it is all for the good. There are
many people who have been doing cool stuff with data that may have been
overlooked for a long time. They are now getting some notice.
In the future, I think we will have a much larger universe of skilled people.
I also think that the education, language, appreciation, and the ability to perceive
opportunities of data is going to move up as well. Particularly in management
side, we will see this increase in knowledge.
The future will also have a clearer understanding where data science is not
always useful. There are plenty of areas where you do not need a data scien-
tist, and do not let anybody tell you otherwise. There is only so much you can
do with data at the end of the day. A lot of what I see people pushing today
are things that cause me to roll my eyes and say, “Okay, fine—whatever.” Just
because something can be done with data does not make it valuable even if it
looks cool. At the end of the day the question always has to be: can you and
are you willing to act on it. If not, you are wasting money and resources on
data porn.
