Database Reference
In-Depth Information
Gutierrez: What was a key lesson you learned from your experience with
the Yankees?
Hu: I think one of the most important lessons I learned was how critical it is
to persuade other people. One of the big challenges of being a data scientist—
that people might not usually think about—is that the results or the insights
you come up with have to make sense and be convincing. The more intelligible
you can make them, the more likely it is that your recommendations will be
put into effect. That is very much something that you have to pay attention to.
You cannot live in a cave and generate all these really cool things that nobody
understands and nobody can put to use.
We had a lot of situations where I would make recommendations that would
sort of go against what conventional wisdom prescribed. In these situations,
I had to make compromises between the data and the real-life applications.
Gutierrez: Tell me more about one of these situations.
Hu: One example is when I was researching batting lineup decisions, trying
to figure out the most efficient lineup for the Yankees. One of my suggestions
was to put a more on-base heavy hitter—a big-name slugger at the time—into
an earlier spot because it would yield more runs. However, this very much
goes against the conventional wisdom of who should go early in the batting
lineup. Conventional wisdom says that fast, base-stealing players who will not
hit into double plays that should go into an earlier spot. A slow-footed slugger
will hit into a lot of double plays, so conventional wisdom said to put him in a
later spot. Also, there was a big concern that he would not accept the move
up in the lineup because he saw himself as a power hitter, which meant hitting
later in the batting lineup. So there were a lot of real-life concerns that I did
not factor in when designing the model.
Eventually, we reached a compromise that involved using a different big hitter
in place of the slugger. This other big hitter was on the roster at the time and
he also had a good on-base percentage. We ended up moving him up into the
second spot in the lineup because he was much more amenable to the move.
I think this was possible because of the culture that he grew up in—where
there was not as much of a focus or stigma based on where you hit in the
lineup. It was a very important lesson in terms of balancing the data with the
people. It is always important to make sure that you keep the people that you
are making recommendations to in mind.
Gutierrez: As you have worked in different organizations such as theYankees,
the US Department of Defense, and now Next Big Sound, what have been key
areas of focus for you?
Hu: Skill acquisition is always on my mind—I am always learning new things.
It is interesting because I feel there are definitely different tracks of skills you
need as a data scientist that all intermingle with each other, but when you are
 
Search WWH ::




Custom Search