Lewis - Graham - Being a great Data Scientist

VIDEO DESCRIPTION:

Play, download and edit the free video Lewis - Graham - Being a great Data Scientist.

In our latest 'Preparing the Unprepared' webinar, Lewis Adams-Dunstan speaks to Graham Morehead, computational linguist, and Principal Research Scientist at Aon about HIRING AND DIRECTING A DATA SCIENCE TEAM.

In the first snippet from the interview, Graham covers...

WHAT DOES IT TAKE TO BE A GREAT DATA SCIENTIST?

Transcription:
Lewis: So I kind of want to move on to the first question Graham, what does it actually take, you mentioned, you know, there's a huge of personalities in a data science team, but what does is it actually take to become a good data scientist?

Graham: Well, you need to be the kind of person who's got your own interests, like your own sense of smell, like certain things smell interesting to you if you're the kind of person who will go down a rabbit hole on YouTube or Khan Academy or whatever to learn the background for something, what you find out is that a lot of the things that we deal with right now have been known about since the foundations of mathematics. Some of the things I use come from the seventeen hundred, the eighteen hundreds, the ancient Greeks. I mean, we have been addressing these same mathematical problems forever. Just now we have better tools and we can do more math faster. And that's all data science is a whole bunch of tiny little math problems, but if you organize those math problems in a certain way, they can learn on their own from the data. So you have to be the kind of personality that goes down these rabbit holes and then what you find out is you will hone in on a specialty and that will become your specialty. And it almost feels like it chooses you for me, its language, and the brain. How does language encode meaning? How does the brain interpret that language when it comes in? That's my love. I've learned more about it every day. I can never stop thinking about it. So every data scientist should have something like that, that really drives you, and you can never get enough of the fundamentals, the discrete math, or the historical statistics. That huge field of statistics, which basically machine learning is applied statistics in some way. A lot of people think I'm just going to go to a code camp and then I'll be a data scientist and you can learn how to use some things other people have done if you do that, it's a good place to start But I feel like I know 10/ 15% of what there is to know in data science and I have been doing this since the late 90s.

Lewis: It's interesting you say that as well, so you're saying that data science is a new thing?

Graham: No, it's not a new thing, we if a new word for it and of course we use computers now, but statistics is what was data science. I mean, Guinness beer is Guinness beer, it's so great because of the student's t test, which is one of the bases of modern statistics. This guy who published under the name student was an employee at Guinness and came up with a way to determine how different two distributions are, and he was just trying to make better beer. We've been doing data science for quite a while.

Lewis: And another thing you touched on is that sometimes your specialism in data science chooses you. If I were a business exec, are you also saying that one data scientist might be different from another, or they all do the same thing?

No, no. They're all a little different. I mean, there are some basic tools every data scientist should have. For now, the common language we all use is Python. So you should know Python and Pi Torch and Tensor flow. You should know the basis of a lot of the current algorithms we use, understand what backpropagation does, what the different neural networks that are popular, and it is really popular, they come and go, but, you know, read the literature, stay up to date on the acronyms that people are using these days. But every data scientist or every, maybe every person should find that one thing that they really feel drawn to, and part of it is just a matter of enjoying your life, but it's more than that if you take two people of the same talent and they both put in the same amount of effort, but one of them enjoys it, the other one doesn't. The one that enjoys it will become better. I think it's because there has been some work that shows that dopamine is released when you enjoy something, and dopamine is involved in memory pathways.

Lewis: So I'm intrigued then, because to become a good data science leader, it doesn't just take being a good scientist.

Graham: You've got to fall in love with the subject, you don't want to be the kind of data scientist who never looks at the data. You're going to fall in love with the data. What are the shapes of the data? So topology is an old field of mathematics, it's been around for a long time, but it turns out topology is one of the most important ones in data science.

Download, play and edit free videos and free audios from Lewis - Graham - Being a great Data Scientist using RedcoolMedia.net web apps