A Blog by Jonathan Low

 

Dec 5, 2016

What Is Making Data Science the Hottest New Career?

There is a vast amount of data out there, and like oil in the 20th century, the economy is becoming more dependent on it.

What there is not is a vast number of people who know how to manage and interpret it so that enterprises may optimally act on it. JL

Laurence Bradford comments in Forbes:

"Data is the new oil." The world is producing more data every year, now, than it did in the entire 20th century. Success over the next decade is going to be largely dependent on how companies are going to be able to turn that data into insights and actually take action on it. Data itself is a commodity.
Gautam Tambay, cofounder and CEO of Springboard, believes that "Data is the new oil." Demand for data scientists has gone nowhere but up in recent years -- and despite what you may think, you don’t need a Ph.D to be successful in the field. In this article, Tambay walks us through the future (and present) of data science.
Why Is Data Science So Hot?
So why is data the new oil? "The world is producing more data every year, now, than it did in the entire 20th century," Tambay says. "Success over the next decade is going to be largely dependent on how companies are going to be able to turn that data into insights and actually take action on it. I think data itself is a commodity. Everybody has it and it’s become cheap to store it and people are sitting on volumes of data, but what do you do with it?"
That’s where data science comes in. In an age where the bombardment of information never seems to cease, most of us just let it all flow over us, remembering a few highlights here and there but quickly moving on to the next thing. Data scientists are the ones who can slice through all the fluff to extract the meaning of all that information: how it’s connected and what its implications are. For companies looking to make strategic decisions, this skill is invaluable.
What Does It Take To Be A Data Scientist?
Most early data scientists came straight from higher education: graduate programs and PhDs. But Tambay attributes this correlation not to the idea that you need higher education to be a data scientist, but rather that the same people who are drawn to academia may be drawn to data science as well, due to their shared focus on "hypothesis-driven experimentation and thinking about the experiments in a very structured way and understanding the statistics."
What this boils down to is that a career in data science doesn’t necessarily have to follow a long academic career; rather, it can replace it. For those who have an academic mindset but don’t want to go into academia and become professors, data science is a lucrative alternative. What does it take? A logical mind. A tenacious attitude. A love of experimentation. A talent for breaking down problems. But the grad degree is optional.
One of Tambay’s favorite real-life examples of degree-free success is Niraj Sheth, who studied liberal arts in college, then worked in journalism, then in marketing, and is now a data scientist at Reddit after teaching himself to code and going through the Springboard program.
Far from considering his background a detriment, Sheth views it as an asset. Data science as a field is a hybrid of several other things," he says. "Fundamentally, it is as much about people -- the users you're building for and the coworkers you're building it with -- as it is about math and engineering. Having a hybrid background myself has definitely helped me understand which parts of data science to leverage at different times."
How To Think Like A Data Scientist
If you’re wondering whether there’s a concrete process to follow, Tambay breaks it down for us into five steps.
"First of all, you want to learn to break down problems into its constituents. Every time you think about why something’s happening, create a hypothesis. This can apply day to day. When you’re doing anything with your friends. When you see something happening, [ask] ‘why did that happen?’
"[Second], think about, ‘what data would I need to prove or disprove this hypothesis?’ Think about why this would happen, think about a hypothesis, think about what data you would need to prove or disprove the hypothesis, then go find the data and see if the data confirms your hypothesis.
"[Third], think about how to bridge the gap between this simple hypothesis-driven thinking to actually running large experiments. That’s where you need to learn the statistics, that’s where you need to think about how to clean and wrangle data, because often data is messy.
"[Fourth], you think about how to organize the data into analyzable form, and that’s when you need the tools, whether it’s Python programming or a language like R or some people will just even use SQL and Excel for smaller problems. But that’s when you need the tools to actually analyze and conduct your analysis.
"Finally, you need tools to visualize and present your insights -- data storytelling."
If you’re on the fence about exploring data science, the first and second steps of this process are things that anyone can do, says Tambay, and they’re a good way to gauge whether you’ll enjoy working in the industry. It all starts, he says, with "asking the right questions."

0 comments:

Post a Comment