From the course: DJ Patil: Ask Me Anything

How do you bring data science into the workplace?

From the course: DJ Patil: Ask Me Anything

Start my 1-month free trial

How do you bring data science into the workplace?

“

(upbeat music) - [Interviewer] Where do we start saying, you know what? Maybe this is too much of a stretch for a data scientist to be able to make predictions like X, Y and Z. - Well, so, I come from a bias world to start. I'm trained in Chaos Theory, which is the idea of small changes lead to giant changes. A butterfly flapping its wings in South America might cause a tornado in Kansas. So, I come from a world where predictions, I'm saying that predictions are hard and unreliable. So, what I think is more interesting is to ask, how do we build new things with data? And part of this gets to, what does it mean to be a data-driven organization? A data-drive organization first captures and acquires data in a responsible way. They then process that data to do two things, they do it in a timely way. They do it to create better efficiencies in business or they use it to turn it into building new products. What's a data product? A data product is oftentimes something where you don't see any of the data. We think of a data product as something where there's lots of data. When you open up the app for the weather, you're going to see maybe a sun cloud. If it's super sunny, it's like super shiny. If it's like, rain cloud and super nasty, it's got like lightning bolts coming out of it. There's no data it's really showing you. But that app is telling you everything you need to know about the world of how to operate for today. You need your umbrella, you don't. If you're a pilot, you're probably going to have a different level of data presented to you, a different kind of level of quality, and so on and so on to different realms. The self-driving car is a data product. It takes data in, it's got to capture it, got to process it, got to turn it into action-oriented things of turn left, turn right, stay on the road, all those questions. How do you make that even more reliable as we get into these questions around, can this thing actually interact within a high-population density? But fundamentally what's happening in these things is this question of, what does it mean to build and make things with data? And how is that, how do we get better at that in every stage of the process? (upbeat music)

Contents