“Data collection is the books of valuable asset a company has” Interview with Ran Oren

Ran Oren

Engineering student at Waterloo, currently working as a Data Scientist at Facebook. He has extensive experience in the past working in the tech industry, specifically in data science roles. He has always been a super quantitatively driven person. He loves algorithms, software, and the future of data and how we can incorporate data.

https://video.wixstatic.com/video/00ea2c_2e50e8380b11442cbc2eabc9932ce89a/720p/mp4/file.mp4

What is your professional title? And where do you work?

Answer: My professional title is data scientist. In the past, I've had a lot of different work experience, do internships and full-time jobs in the tech industry, specifically in data science roles. Like this past summer, I was working full-time at Facebook. Now I'm going to go back after I finished a few courses to continue working for them. In all these roles, we have worked with massive amounts of data to drive the business. At Facebook, we would use data to make our products better. In the past, I've worked at a company like wish, which is another tech company, that would use data to help make marketing decisions, and all of the algorithmic systems that we have in place.

What does being a Data Scientist at Facebook look like?

Answer: Every company has their own job roles and requirements, that's expected of a data scientist. At Facebook, it's definitely more on driving product decisions. So the role as data scientists would be you're using the massive amounts of data that you have to drive the direction of a specific product. If you're working on Instagram, you might use Instagram’s data to decide which features to build out, or how to move forward with a product given user behaviour. So it's really just about using data to drive important decisions.

What kind of projects have you worked on?

Answer: At Facebook specifically, I was working on their dating platform. We were essentially using a lot of data we had on how dates interact with each other, all the different kind of elements and how users interact with the user interface to just make that product better. Like I mentioned, in the past I worked at wish, there, I had built a machine learning model, which is like a predictive model that essentially looks at or predicts which users will make transactions in the future. There's a lot of data that goes into building a model like that, and as a company, they would use that model to figure out, what products to recommend to certain users? What kind of advertising should feed to certain users, etc. So a lot of very data intensive products I've worked on.

How would you explain Data collection?

Answer: Data collection is an integral part of this of data science and just software in general, I'd say I would describe it as the means in which you get information into a structured software database. So there is data all around the world, that it's not possible to collect all of it. But data collection is the process of there's a physical thing, or something tangible has happened, how do we actually record that? And how do we keep on recording that? It's an entire field in itself. But for example, most tech companies, they have the luxury of user logs, so every interaction you do on a mobile app or on your computer, it gets tracked and it gets logged. And that's just one method of collecting data.

How is data collection executed?

Answer: I think it totally depends on the use case, like what problem or what you're trying to collect. As I mentioned, for a technology company, there are logs, they're able to track how you interact with that mobile application, web app, and they can use these data to predict future behaviour and stuff you can do. For other examples, things like sensors, if you want to track how many cars let's say are passing through a highway every day. You can put a sensor on there too, and that's one way of collecting data. In marketing applications, things like QR codes are ways of tracking how users are going through a certain process or interpreting marketing material. There is so many different ways you can collect data, it really depends on what specifically you're trying to collect.

Why is it important to integrate data collection?

Answer: It's important integrated a collection, because without that component, there is no data analysis, there's no analytics, all of these futuristic modern things that businesses should incorporate would not be possible. Data collection is the source, it's the capital, it's the books of valuable asset a company has, and you can't get data without collecting that data somehow. The years 2021 when we're doing this interview right now, companies out there they need data to start, and how they get that data is just absolutely critical. It's what feeds all of their other initiatives, software related, business related, etc.

In the graphics communications industry, how could you see data collection being implemented?

Answer: I think the graphics communication industries is really interesting to focus on because how you collect data is a little bit more complicated. So as I mentioned, if you have things like, tangible printed, advertising or just any kind of tangible material that a person is interacting with, you need different ways to track how that user is interacting with that material. Like what I previously mentioned, things like QR codes, if there's a printed piece of material, and you just embed it or you include a QR code on that material, and I, the user can scan that, that opens up so many doors, because now you're able to make a connection between a printed ad, and who actually used that printed ad, whether or not it actually worked. I think something like a QR code, it is gaining popularity as people realize how good of a data collection mechanism that is. But I would say, QR codes is definitely a critical part of data collection in the graphic communications industry.

What kind of problems could we run into, when implementing data collection? And what potential solutions could be implemented?

Answer: I would say, even with a QR codes, they will not give you a perfect picture of what actually happened. So let's say I encounter an advertisement, and it'll tell me to go buy this product. Not every single person that runs through this ad will actually use that QR code. So when you go ahead and assess how that campaign did, or how that piece of material did, it's hard, because you're not necessarily getting a perfect picture of the exact amount of people that actually, went on and bought that product. As not everyone will actually end up scanning that piece of material. So it's just the data imperfection and not be able to get a perfect picture of the situation, that is an inevitable reality data collection. But there are statistical methods you can use to go around that. So we could assume a specific percentage, or you can add a certain percentage point, there's a lot of different kind of technical things you can do.

Relative to Data and A.I, what does the future look like for all businesses?

Answer: I think that question goes back to what I was mentioning before. Data and AI, they're all very related. I just think that businesses will be not forced, but they will naturally have to start getting involved with data collection with statistical modeling, just because everything in the world is becoming increasingly tracked and monitored. Computers can process things at a rate in which humans cannot. If you're a business and you need to make business decisions, you need to make automated marketing decisions, you need to make all these different decisions. Computers can do that better than humans, and what drives computers to do that is data. So I think the future for businesses are just like those web 2.0 where, every single business needs a website, no business does not have a website. I think the next iteration of that will be, businesses need data, just like you need a website and how it's like you got to start collecting data to stay relevant.