x
Advertisement
Scroll to Continue

Most Important Trait For a Data Scientist? Curiosity

A Q&A with Ooyala's Matt Pasienski

By Published on .

Credit: Matt Pasienski, right

There's no shortage of people watching video online and certainly no shortage of data detailing that video viewing. It's Ooyala's job not only to manage and deliver digital video, but to help media firms understand who their viewers are, and give them what they're looking for based on up-to-the-minute information.

As the company's chief data scientist for two years, Matt Pasienski spends his time turning "crazy messy data sets into something meaningful for our customers." But the sheer volume and speed with which data rushes towards him and his team means he needs more than his PhD in physics to make sense of it all. For Mr. Pasienski, the most important trait he or anyone else in his line of work needs is curiosity.

"I'm insanely inquisitive, which perhaps more than my education makes me qualified for this job," he said.

Toting a PhD in physics from University of Illinois -- "with emphasis on quantum information, phase transitions and cold atoms" -- Mr. Pasienski is not unlike other top data crunchers with backgrounds across the sciences from biology to mechanical engineering to, yes, physics.

He got his undergrad at University of California Berkeley in applied mathematics and physics.

Optimizing video content to best suit what people are looking for is a key element of what Ooyala does for clients such as Fox Sports and Vice, but measurement is critical. While things like the amount of time someone watches a video or interaction rates are readily quantifiable, said Mr. Pasienski, measuring engagement remains a challenge.

For instance, how should mid-roll ad viewing engagement be gauged in comparison to pre-roll? Mr. Pasienski suggests mid-roll viewing is "high-intensity" and expects there to be a more nuanced way to quantify that "just like a ratings point or something like that," in the next few years.

Ad Age: What educational fields of study and professional backgrounds help develop the best data people?

Mr. Pasienski: I believe that anyone can do this job, if you have one trait -- intense curiosity. OK, you need strong mathematical and computational skills, but the most successful data people are the ones that have a desire to keep looking. (Most of us love trivia too.) If I were hiring someone, my question wouldn't be about his or her mathematical background, I would ask them what the most interesting thing was that they learned last week. If there response is great, and it's obvious their inquisitive by nature, that's the data person you want at your company.

Ad Age: Do marketers get what data scientists do?

Mr. Pasienski: I wish marketers would understand that they need to be data scientists themselves. They need to speak data and they need to understand data. Only then will they be able to make informed, or data-driven decisions and get the most out of their marketing spend and build better products. I also think marketers are missing opportunities to include hard data in their communications to increase the authority and relevance of their message. A really good graph or stats based on real measurements will help marketers get above the noise.

Ad Age: What could data science people do better as they navigate the world of marketing?

Mr. Pasienski: Slimming down the message. Most data scientists focus on facts and figures, but the challenge operating in the marketing world is not just mining data for insights, but knowing how to communicate your findings. It's your ability to distill down the data and figure out the pieces you need to tell your story that'll be valuable to marketers. It's a process. You might make 100 graphs before you find the right one that will tell the story in a simple way, but that's where data delivers the most value.

Ad Age: The FTC just held talks to discuss privacy approaches for The Internet of Things. Do you think it's a good time for government to regulate track-able products (the data-generating refrigerator is a common example) and how companies gather and use data from them?

Mr. Pasienski: Government should be an innovator and a standards setter as opposed to a regulator. The lack of consensus actually can really hurt companies involved in the space. Whenever you're starting up a big system with a lot of different people in it, it can take a long time to develop consensus, and when it takes a long time for everyone to get on the same page, it can have really bad effects. There's definitely a role for creating trust and goodwill, as long as it doesn't take the form of really draconian and ham-handed enforcement.

Ad Age: Work with any cool datasets lately?

Mr. Pasienski: I just finished looking at when people use tablets by time of day (it turned out there was a really big spike in the evening). It was a project where I went in thinking one thing and came out proving another. Those are the most interesting projects for me. The best part of my job is finding things that you wouldn't expect, where the only way to find them is by going and looking. It's the process that I enjoy the most, stumbling upon all kinds of cool stuff by being persistent and constantly digging.

Comments (0)