This is a cityscape

Executive Panel: How Can Data Science, ML, and AI Best Support Executive Goals

Today is a special episode. We welcome three executive guests from different organizations to share their experiences and insights about how data science can best support executive goals.

Ginette Methot: I’m Ginette

Curtis Seare: And I’m Curtis

Ginette: And you are listening to Data Crunch

Curtis: A podcast about how applied data science, machine learning and artificial intelligence are changing the world.

Ginette: Data Crunch is produced by the Data Crunch Corporation, an analytics training and consulting company. There’s a lot going on here at Data Crunch. Just this last week we finalized the merger of Vault Analytics and Lightpost Analytics under the new banner of the Data Crunch Corporation, which improves our capabilities to serve our clients head over to datacrunchcorp.com to check out our training and consulting offerings.

For our executive panel, today we’ll be talking to Simon Lee, the chief analytics officer from Waiter; Fatma Kocer, who is the vice president of data science engineering at Altair, and Rollen Roberson who is the president at Trianz.

Curtis: So, welcome everyone to the executive panel. We are super excited to have you guys here. You are all executives and companies that are doing amazing things with data science. So the audience knows, again, we’re talking about today, the topic is how data science, machine learning and AI can best support executive strategy and business goals. How, how does that function really work? Let’s start maybe with Simon and then Fatma and then Rollen, if you could just give us a little introduction, and we’ll get going from there.

Simon Lee: Thanks. I’m Simon Lee. I’m actually kind of a mixed bag when it comes to data science and analytics. I’ve got about 20 years of experience using analytics and advanced algorithms, you know, in a whole bunch of different industries like transportation for example, airline rail, trucking, ocean carriers, printing, publishing, manufacturing, finance and delivery. Delivery is where I’m currently at. Waiter is a restaurant, food delivery company in small and mid size market. So probably a lot of people haven’t heard of us because we’re in the smaller communities, but, we’re trying to make a big splash. So yeah, that’s who I am.

Curtis: Awesome. Thanks for being here.

Fatma Kocer: Hi, this is Fatma Kocer from Altair engineering. I am a civil engineer by training, although I never get a chance to practice it. Um, my background is multidisciplinary design, exploration and optimization. And I was in the auto industry before I joined Altair. Um, there, I’ve done several things throughout the 14 years that I’ve been here, but always keeping, designing solution optimization as the core of my responsibilities. And Altair is a global technology company. We provide software in solutions for product development, data intelligence and high performance computing. We are located at headquarters in Michigan in Troy, Michigan where I’m speaking from and we have offices in I think 25 countries now. So that would be me.

Curtis: Great. Thanks for being here, Fatma, and, ah, Rollen.

Rollen Roberson: Right. Thank you. Good Morning. Rollen Roberson with Trianz. You know, for my own background, I’ve a similar to Simon. I’m kind of a mixed bag, I’ve been in the industry for 20 plus years, I’m solely in the digital transformation space. Uh, working from startups, mid-level companies through global service integrators, uh, working with Trianz currently to really expand the growth and a use within AI and IoT within the organization. And our customer base, Trianz is a company that has 1,500 plus employees, global offices mainly serving the, upper, mid-tier and enterprise level customer base, uh, solely focused on digital transformation and the use of those higher technologies for greater return on value.

How Can Data Science and AI Have an Impact On Your Business

Curtis: That’s awesome. So lots of experience here. Really great panel. We’re excited to hear what you guys know and can offer to the audience. So we’re going to jump in right here. And the first thing that I really wanted to dig into is as an executive in your specific industries, how do you approach leveraging data science and machine learning and AI to have an impact on your business? What’s your focus and how do you make that work in your companies? And maybe we can start with Fatma here and then after she’s done and as she’s talking, if other people have comments, please join in.

Fatma: Thanks, Curtis. I think the biggest impact that data science, machine learning, um, has on our business and on our customer’s business is to enable and speed up innovation. The way I think about data science is Data science allows us to think in the multidimensional way. The problems work, the applications work, and we no longer have to think linearly because it’s simplified seeing the multidimensional aspect of your problem. You know, this gives us relations between changes in the system and the corresponding changes in the system performance. Whether the system is an engineering system or it’s a financial system, it allows us to build predictive models using machine learning. And of course we know, we all know that once you have predictive models, you can do prescriptive analytics, explore many alternatives, the trade offs, you know, expose new ideas and new products that paves the, the way to innovation, we have many customers throughout the years that has, um, that we wrote with to, to deploy this. Um, it has been mostly in the engineering domain, but recently with our acquisition of the data intelligence company, we are also leveraging these technologies for business cases. For example, we have a customer in, a large consumer products brand. The way we use data science to find optimal package designed to use material waste, which is an engineering problem. And at the same time we do optimal traits spend to increase the sales and maximize the revenues, which is a business problem.

Curtis: Got It. I think there’s a lot more we could dig into there but let’s give the other panelists an opportunity to respond as well.

Simon: Yeah, I love Fatma’s answer. It kind of mirrors my own experience here at Waitr. What I really love about Waitr is just the multidimensional and multi kind of the different aspects of data science and analytics in general, for example, forecasting. What the order volume will be, what happens on holidays, unusual events, Superbowl, parades, et cetera. What happens to the order volume? We need to schedule that. And then, in terms of just optimization, how do we get an efficient schedule out? Cause we have thousands of drivers that we need to schedule. And how do we match the supply and demand once we get the forecast, how do we, when we’re actually doing the operations, how do we match the drivers to the orders? Again, another optimization problem. And then, and then in terms of machine learning, what can we learn about the three pillars of our business? The customers, the drivers and the restaurants. How do we classify them? What do we know about churn and their lifetime value and their preferences and all that sort of stuff has tremendous business value. And it’s just an exciting time and an exciting business where we can apply all these different aspects of analytics to make a meaningful impact. Yeah, I would agree with that.

Rollen: You know, as we look across our customer base, you know, we’re data science, data engineering, artificial intelligence, machine learning. All the technologies that we’re dealing with today helps us take and make the,the difficult easy for our customers. We find many of our customers now are quite frankly struggling just to keep up with the amount of data, the different types of changes in the environment that they’re dealing with on a day to day basis. And what we do is actually make those type of challenges easier to overcome. We look at the tsunami of data that they’re now getting presented with and we take that and present it to them in a way that is not only easier to use but consumable. Thus they can actually make a better business decisions with the type of data now, the right information that’s presented to them so they can make a better day to day choices.

Curtis: That’s a great overview. I’m curious now, now that we’ve talked about these various use cases across all of your industries, actually, what are some specific concrete examples that you guys have seen where you can help help our audience really get a grasp on, okay, here’s a specific use case that we saw and the results from it and kind of how we built that so that they can see this process.

Rollen: Sure. Be happy to. So as an example, we’ve recently been working with a large global logistics company and part of their transformation internally is to actually migrate from human driven vehicles to automated vehicles for types of shipping and logistics. Now part one of that transformation is just getting their arms around the telemetry and the telematics that’s happening within the vehicles on how a human driver interacts day to day. Now as part of that transformation, it’s actually the, not only the sensors that had been upgraded around the vehicles to collect the information, but actually the collection then and distribution of that data to understand what’s happening. So real time use case is that while we might understand how to move a driver from A to B, what you’re really seeing though is, and the intent was maybe I can move them to a location, and see if a storm is in a way to maintain my logistics schedule.

Rollen: But with some of that sensor data, we’re now collecting weather information across all the different vehicles nationally. And so with that we understood that now we have a better weather collection database than the national weather service, it is even actually more real time. And just having that data is valuable to itself and actually can be considered a revenue stream. So we identified new revenue streams for our business that didn’t even have that in mind when they first went down this journey and now we started open up new ideas and new ways to collect revenue that they never knew that they had before. So it’s not just attacking the problems that they’re, that they’re interacting with today and trying to move them forward, but it’s, it’s identifying new sources of revenue that they never knew that it was even capable of having.

Curtis: Got It. I love that. Because revenue and profit are what keeps us in business. Are there examples that people have?

Fatma: I have a more generic, response to it. As, you know, what makes success is, you know, identifying the right case in one that has business value, but technologically within reach. And I think the example that Rollen gave is a perfect example of how all those pieces are aligned.

Simon: And, I find one of the most rewarding things in terms of the work that we do is, you know, as human beings we tend to be fairly proud of our ability to make good decisions. And I think that as well people are working in online or manufacturer or even executives, we tend to be fairly proud. And then when somebody comes along and says, you know, there may be an algorithm or an approach where a data-driven method that can help make a better decision, the immediate response is usually skepticism. You can’t figure out all the, all the different factors. I, I’ve been doing this for, you know, a long, long time, and when you do deliver something that actually far exceeds what a human can do, sometimes you get this incredible change of attitude and perception and they say, yeah, that is actually better than what I can do.

Simon: And that’s actually very rewarding to see that transformation to say, you know, we can help make better decisions with this massive information that we’re collecting. And we’ve seen it in the airline industry, we’ve seen it in printing and publishing where and in printing and publishing, we, there’s a schedule of how to, how to construct the printed material, which is a fairly complicated thing, but we represented it graphically for one of the experts in the field. And when I showed what he did and what the algorithm did, he actually stopped and said, oh my goodness, that’s absolutely beautiful. And just to get that success, which translates into dollars is quite rewarding from a personal level because it’s, you can, you can actually see, yes, we’re making a difference here. Even the experts in the area see that and acknowledge that. And I think that’s really, really rewarding.

What Are The Biggest Challenges to Implementing Data Science and AI in Your Company?

Curtis: So lots of use cases here and lots of transformation as we’re talking about this. Simon, you brought this up just now that there’s sort of a hesitancy for people to accept maybe that an algorithm can do better than I can do right? When once you show them that there’s a change of mind, what other areas are you guys seeing that are big hurdles or big challenges that an executive needs to be aware of when they’re trying to take advantage of what the capabilities are nowadays? What do they need to be thinking about? What are the pitfalls?

Rollen: I’ll give you an example. You know in many cases what we find with our customers is that they already have a significant amount of data over the last a decade. They’ve been struggling quite a bit with how to leverage that data effectively for day to day use. And you can acquaint it more with actually the customers really is in firefighting mode because they are actually taking bits and pieces and trying to stitch a relevant story together as to what that data really means. And then they ultimately find that they don’t have a complete view of what that data is. So they have to go find out more. So time and time again, they’re fighting, struggling, trying to make some coherency of the data and bring meaning to it. Now with a lot of them, more recent advances, I mean, and you’re getting more and more data now that’s being moved not only from customer data that’s being brought into larger scale ERP systems, but you’re also getting IoT enabled devices.

So this wave of data is now going to completely swap, you know, the customers that are there. So some of the biggest changes is getting the customers to understand that they can move past this firefighting mode and actually use a lot of data science techniques, visualization, machine learning in order to augment the data. So they don’t have to look at the bits and pieces of the data, let the machine learning algorithms do it for them and then they can step back and actually take a broader view of the data to really see the patterns and trends that are happening and make use of that.

Fatma: So from the engineering data science perspective, the biggest challenge is actually availability of data. And it’s not that we don’t create data, but it’s that we don’t usually, you’re not very disciplined about this data. We don’t save the data because they take a lot of storage space and usually think that if they say, you know, once we find a good design we just delete the rest of the data. If the are semi-disciplined and we keep them, we store them some of them are not as usable for a machine learning algorithm to be able to crunch them right away so they need to be formulated with meta data. So in other industries I think there is data but usability is still an issue as Rollen also mentioned. So this creating a data fabric of getting data from all types of forces, prepping existing, uh, collaborating, visualizing, doing predictive analytics can be challenging in any application.

So we have data intelligence solutions that fits into this notion of all with the, the data prep, high end predictive prescriptive analytics and visualization in a robust automated environment. One example is for, you know, we usually go to customers to discuss machine learning applications. And the first question we ask is, you know, where is your data, how much do you have, and what are the formats? And most often the answer we get is we don’t really save data, we delete them, and the data that they save is usually very minimal. We can go back to them and say, well you saved some metadata from all these large files, in pdf, in reports, so you can actually get your data back from the saved sources, from the pdf files and automate the process so you don’t lose all the data you think you lost by deleting the raw data.

Simon: Maybe from an executive perspective. I think one of the challenges that I’ve seen as the difference between the acknowledgement that data science can have great benefits versus the conviction that data science can have no great benefits. So a lot of executives kind of have this understanding that yes, for maybe some companies for some applications, data science can do something but you know, maybe not in my company or my industry versus the conviction where yes, we have this huge amount of data that for example, Rollen is bringing forth with a IoT and consumer data, huge amount of data. And honestly we don’t know what benefits we can get from the data from a human perspective, we often don’t know the interactions and relationships found in the data that machine learning can bring forth. And to say that there’s a discovery process that actually will yield tremendous benefit, even though we don’t know a priori what that is, you kind of need that conviction to kind of exchange between acknowledgement to say, okay, yeah, some people can do it and it’s done a lot of goods in some places to that conviction that says, yes. In my industry, in my business there’s a world out there that I’m unaware of, but it can yield tremendous benefit.

Rollen: Yeah. If I could add onto what Simon was saying, you know, he’s correct in that you do have to have as a customer, a consumer of the data, you do have to have the conviction and it’s almost a, an inherent sense that you have this world in front of you and you’re not quite clear whenever you’re looking at it. And you know, another use case that reminds me of this is we’re currently working with the legal firm and they have a vast amount of data about the cases that they work and they understand that they currently bill and generate revenue in a certain way, but they have an inherent feeling that, you know, there might be a better way of doing this. And we’ve employed data scientists and different algorithm techniques in reviewing their data just for discovery exercise to try to uncover, is there a different way to categorize the cases, a different way to look at how they’re executing trials? A different way to look at how they’re reviewing evidence. Um, these types of things. You know, it’s very time consuming and this type of data is not readily or easily reviewed and you can only do it through machine learning techniques. So by doing this, it’s really a discovery exercise to try to find out new ways of doing business. And you have to have the thought, the thought leadership in moving forward to say, I know there’s something there, we just have to find it.

What Are The Biggest Opportunities for Data Science and AI That I Should Take Advantage Of?

Curtis: Got It. And so, so would you guys say in terms of biggest opportunities or what executives should pursue is this idea that we’ve been discussing of just having the vision and the belief that I should try this and see what comes out of the data? Is that maybe the biggest opportunity or thing that executives need to do? Or are there some other things to consider?

Rollen: Well, I think the discovery exercise is not necessarily the first step in all cases. I think from an executive position, you’re typically looking at what’s the best return on investment that I’m going to have by applying this type of a technology or a process or procedures and discovery may or may not. That’s an experiment and you have to have a little willpower in order to pursue that. But as you’re looking at the other challenges that businesses have, usually we try to find very pinpoint solutions that we can apply these types of techniques to, to provide the type of return that the client executives are really looking for. They need to make sure that they’re seeing a return on the back end of this type of exercise. And by doing that, a lot of cases, it is through some of the more fundamental work just to get your arms around the issues. Whether that is, how do I consume the data, the collection of the data, that cleanup of it, or even the migration of that data to make it easily moveable from a location to another. So I can use it more effectively. Just those preparation activities, add a lot of value and what it’s they need to do so later they can apply some of the higher end processes.

Curtis: Rollen, you made a good point. There are certain things you can do to get value faster than maybe a discovery process. What’s the typical, and maybe this is too broad, but what should someone expect as the timeline of, I start down this road of using data science or machine learning, to where I can get my first return or, or a quick win or something. How long will it take?

Simon: It kind of depends on your maturity to begin with in terms of, you know, Fatma was describing scenarios where the data hasn’t been stored. So, you know, how do you handle your data and then what if you’re not even storing the data. It’s really hard obviously to make any progress. But you know, assuming you’ve collected a bunch of data, et cetera, you know, I guess the first thing people often do is they just report on the data. Are you, is your data put together well enough so that you can simply report what happened in the past? That’s one level of maturity. And then another one would be given the stuff that you have, are you able to predict what’s going to happen in the future, you know, forecasting, etc. Based on past data, that requires a certain level of expertise as well. And probably if you’re able to report, you’re probably able to forecast if you have some level of maturity in terms of your data science. And then what Fatma was also describing was prescriptive analytics. How do we make the best decisions based on the data that we have right now? Can you apply the data to actually make good decisions? And I think that’s yet another level of maturity that probably takes, you know, and in the order of, depending on their size of team, maybe months, maybe a year or two to in order to come up with machine driven decision making. And that’s, broadly speaking, that’s kind of the three levels of maturity at least as I see.

Fatma: Yeah. At the risk of repeating what Simon said about the timeline, and the investment that it takes to do machine learning to get value from data science and machine learning, I agree. The data engineering part, that actually is the most involved and the one that takes the most time collecting the historical data, augmenting the historical data when needed. Once you have data in a usable format, I think the machine learning data science part is an iterative process because you learn as you go through it and it’s not the time consuming process because you are learning as you are exploring, and then you’re doing prescriptive analytics running what if scenarios, but data pipelining, data engineering definitely, whether it’s sensor data that’s coming from the field or it’s running simulations, you know, you have simulations that take in, you know, couple of hundred calls or nights a day, for a couple of days. So collecting enough data from these sources is challenging and a time consuming part. So if you have data on doing the rest of the work, the data science work is the fun part and the value added par. Getting that to that level of data discipline and data security is the challenging part.

How To Cut Through the Hype and Keep Up With Real Capabilities

Curtis: I want to maybe switch tacks a little bit cause there’s a couple other points that I think our audience may have questions about and one of them is the technology behind and what’s driving machine learning and artificial intelligence. Data science is moving so fast and there’s so much hype in the media. How do you suggest that executives who aren’t necessarily really deep in this field, how do they really keep up with it and know what’s real and what’s not, what the limitations are, what the opportunities are, what sources do they look to? How do they, how do they handle this?

Simon: Um, the, the way I do it is basically networking, talking to my colleagues, talking to, you know, this panel for example. It’s just understanding the types of problems that people are working on and the ways they’re using the techniques are using to solve it. So that way you kind of can see where the entire industry is going from a practitioner’s point of view as opposed to the media and the hype and whatnot. The other thing that I think is, is really helpful, at least for me personally is I still do analytics. I still code. I still, I’m still involved at a very low level. On occasion. I can’t do it all the time, but that helps me to stay grounded, to understand, yes, these are the tools that are available now. These are the techniques that people are using, at least so that I have a hands on understanding of, and I can’t have a handle on everything, but at least a portion of what’s happening in the analytics world.

Fatma: And then the engineering . . . sorry, go ahead Rollen.

Rollen: Well, I was just going to mention quickly that I think for the layman, an individual who is not historically within data science, their own industry organizations, whether they’re from insurance, logistics, wherever that might be, medical, their own industry, um, organizations that are really not technology focused but yet still highlight advances and what other companies and what their peers are doing in the industry is very important. That’s a very good link for them to have to understand what’s happening in this type of area. And it doesn’t have to be really these spectacular leaps, you know, for them to understand about how to take advantage of this. Either the data science capabilities that are out there, the ability to leverage this can be done in bite size pieces that are still very valuable to them because this can be an overwhelming capability for them sometimes, if they’re not in this field to understand it, and as stated earlier, they’re moving very rapidly. If you can take the opportunity to understand what others in your field are doing and others in your industry, you can take it in bite size pieces and this will shorten the time that you have for adoption of some of these newer technologies and practices. But as well, still get a leg up on some of the challenges they’re experiencing day to day .

Fatma: In the engineering world, I feel like if you’re not out there building mean robots, you’re not considered to be doing AI even though you know, we may be able to tell that your part is going to fail in two weeks and then you should schedule your maintenance and repair part, you know, accordingly, which reduces a lot of down times, it’s a little of money and possibly save some risky failures. It doesn’t sound as impressive. So avoiding the hype is difficult, but I agree with the other speakers here that it’s our responsibility as data scientist to show others the possibilities without hyping it. Being able to explain the value and then also to point the challenges without being discouraging.

How Do You Handle the Data Science Talent Shortage?

Curtis: Something else that comes up a lot as I’m talking to people is the talent shortage. Some people in certain industries are having a bigger challenge with this than other industries, but generally people find it difficult to hire the kinds of people that can do this work and do it well, and in a way that produces value for the business. So I’m curious, what is your advice for confronting this talent shortage? How do you hire and structure data teams so that they can succeed for the business? And how do you approach that?

Rollen: Well, I think there’s some basic blocking and tackling an organization needs to have internally in order to make it effective. Most organizations have, if they have their own it organization, have a data practice someplace, whether it’s for governance or just management of this type of storage that they have internally. But I think it’s inefficient for companies to have, unless you’re in the field of data science to have, you know, a number of data scientist as an example, typically just to have access to one or two that might be relevant to the size of your organization is important. But the amount of data engineers, the type of analysts that you need, and then the tools and structures that are in place really is what provides the horsepower an organization needs. Because while data scientists can do some of this type of work as well, there are different skill sets that need to be leveraged. So it’s usually just applying the right people at the right time and the right place. But as you stated, a lot of these skill sets are in high demand within the industry today and some of them come at a very premium cost. But the juxtaposition to that is, is the experience that you need to have as well, in this field, and that’s very important, and sometimes that’s what makes really finding the right people at the right time to difficult proposition.

Fatma: Yeah. I don’t think there is a shortage of data scientists but there is definitely a shortage of domain experts who knows data science. So there’s a lot of interest in studying data science, but I really wish people didn’t forget to leverage strengths, their strengths in the domains that they have been working on until then. So we usually identify people that know the domains that you’re interested in. And then among those we pick the ones that have demonstrated an understanding and motivation to work on data science. And I feel like that worked really well for Altair so far because we have a pretty talented data science team, that deals with projects, develops processes, and looks at our own tools to see how we can improve them using data science.

Simon: I think it’s, it’s difficult, you know, depending on your industry and your company to compete with the big names out there that are inflating the salaries, cause it’s a supply and demand problem. And so the way we’ve approached it, or I’ve approached it, is to use some maybe non traditional hiring, hiring people in quantitative areas, not necessarily data scientists per se, but experimental physicists for example, or mathematicians. I’m getting people that have a very strong quantitative background. And the trait that I look for the most is intellectual curiosity. Are you interested in solving problems? Are you interested in understanding the world, understanding data, understanding business? Cause if you have that, then, honestly, the tools for data science are getting better and better and easier and easier to use. Different companies are rolling out new products every day. But you kind of need that basic quantitative understanding and intellectual curiosity to help you succeed in data science and probably almost whatever things that you were pursuing.

What Do The Next 5 to 10 Years Hold?

Curtis: As this industry continues to mature, and the tools are getting better and better as you say, what do you guys see happening in the next maybe five, ten years in terms of the capabilities that we’ll have, how that will shift how we structure data teams or hire people and just in the capabilities of what we’re able to produce for companies. So what do you see? What should we be looking forward to and planning for?

Rollen: I think Simon actually made a good point in his earlier comment where the advancement of the, a lot of the tools, at least that I’m familiar with and working with today is really around the collection and the visualization of that data. So the tool sets available to data scientists and data engineers today have advanced so much and I expect that to continue in the next few years. Whereas now as more and more data is being presented, it’s still going to stretch the limitation of the current tool sets and they’re going to need another level, another layer, of the ability to understand that data, to work with that data, and to mine it effectively. So, as Fatma was actually talking about earlier, the ability to leverage a lot of those machine learning techniques to actually helping push us further along. So another level of aggregation of that data I think is where we’re heading.

Fatma: I agree. Preparation and visualization are the most important pieces that I think the industry needs to focus on right now. Because the day to day users are going to be interacting with machine learning through preparation and visualization. They’re going to let the data scientists take care of the modeling part for their applications. And I think as data scientists, we need to realize around the standards, the day to day user, the analyst, that they have your own fires to fight and they have their own to do list. So you have to provide them with ways to use these predictive models in a repeatable robust way.

Simon: And I think what’s exciting about this area is, if I were to predict what’s gonna happen in the next five years, I’d probably be wrong. Because I really have no idea. What I do know is there’s a whole bunch of really, really smart people and really creative people in this discipline that are trying to solve problems that have never been solved before with different approaches and different algorithms and different tools and different technologies. And I don’t know what’s going to happen. And I think that’s what makes it exciting to see, oh, we can do that. Oh, you know, a long time ago we would’ve never thought that a machine could understand natural language. It’s in your living room or it’s in Alexa or whatever. You can talk into your phone and it understands what you’re saying. Twenty years ago we’d never be able to predict that. And now just the applications and algorithms and all these different tools that, that now are commonplace, you can download a library that can do so much now that was just unthinkable even 10 years ago. And I think that’s what really makes it exciting and really makes it a fun, energizing, place to work. I really don’t know what’s going to happen in five years, but I’m really excited to see what does happen.

Fatma: I very much agree with that sentiment. One of our focus areas is 3D shape recognition and the amount of developments in that niche domain itself is just amazing.

With Advancing Technology, Will Data Science Be Commoditized?

Curtis: I’m curious now that you’re bringing this out of the, these methods and things are getting better and better. Tell me your thoughts on this, but it seems like there’s a little bit of a difference between applying methods that are well understood to common business problems, things like this. Maybe some of the more traditional machine learning methods and these kinds of things versus the cutting edge. If I’m going to build an autonomous car with deep learning and all these systems that seems like even now a little bit of a different skill set and a different way to attack business problems. Do you guys see sort of the commoditization of some of these methods to where it’s going to just be easier and easier and maybe you won’t need the data scientist role in the future? Or how do you see that shaking out?

Rollen: Well, I do think that there is a commoditization that’s happening, so I do think it’s going to be more consumer friendly as you move forward. But I don’t see that there’s gonna be a relaxation on the need for data scientists moving forward. You know, one of the great things that the other panel members are saying is, you never know what’s going to happen. And as we move farther and farther and become more aggressive, and a lot of this on the forward edge of technology, does let you discover so much more and it’s those undiscovered nuggets that really lead to such great innovations. But all the other things you’ve learned along the way, you never know what’s going to spark and become relevant in other fields. So it does become more commoditized and consumer friendly as you move forward. And that’s some of the great things about how people take all these little bits and put them together in such different ways that keeps it exciting.

Simon: I also think that it’s progressive in this sense. Certain things will become a commodity, but what that will enable you to do is solve harder and harder problems, problems that we would consider impossible today because of the toolset and, things that we can build upon, in the future might become sort of the cutting edge, and they’ll be able to solve them in the future. So I think there will never be a lack of problems. We’ll always have problems, but the problems will become harder and harder and richer and richer. And I think that’s why maybe it won’t be called data science in the future, but will be some discipline that builds upon the techniques and the technologies that we’ve kind of mastered today and that will be trivial later. It was really hard now, but it’ll be trivial later.

Curtis: This has been really awesome. A lot of great comments, and really good insight here. I just want to leave the last couple of minutes that we have to you guys to have the last word. We’ve asked some questions that I’ve heard from people we’ve talked to, that they’re thinking about and problems that they’re trying to solve. But I want to leave the last word with you guys, if there are other things that we haven’t touched upon that you feel are important or that you’ve been thinking about that you think are valuable.

Rollen: Well, I’d make the comment that from an end user or consumer point of view, don’t let this scare you. You know, while you might be facing a lot of large challenges, large data sets or working with other companies, providers and, or even your internal data scientists, don’t let the topic become overwhelming. You have the resources available to you so that you can learn more about it, become educated on how to take advantage of it. And you can take what seems to be an overwhelming problem and make it, split it up into bite size chunks that are adaptable to you, consumable and allow you to actually execute upon something. And to get the best return. It doesn’t have to be scary. And the, there’s lots of techniques and uh, you know, people out there willing to help you do that.

Fatma: Yeah. I think my last words would be similar. Start collecting data if you haven’t done it yet and start experimenting with it either on your own, or ask for help. There are many companies like ourselves that work on data intelligence. If you haven’t yet, this is the right time to start looking at how you can extract value from your data, and you’ll be amazed to see how things that you consider as a challenge or an issue actually have solutions on in the field.

Simon: Yeah, and I would echo that to say, wherever you are, your company, wherever you are in the journey, I think it’s exciting and rewarding at every stage. Whether it’s just collecting data and getting the most basic insights to super sophisticated machine learning. It’s exciting and rewarding and I think it offers potentially great returns. And I think that’s why it make is kind of an exciting thing to consider. And I think Rollen made a great point. Don’t be scared. It’s not magic, but the results it can produce are quite magical. There are humans that understand it to some degree and so it’s not a scary thing. It’s a journey that will take some time and wherever you are in terms of maturity, you will be able to see results.

Curtis: That’s great. Well, thank you guys. Again, these have been great insights. On behalf of our entire audience, I just want to say thank you for your time. I know it’s valuable and it’s has been really great talking to you all and I appreciate it.

Rollen: Yeah, my pleasure.

Fatma: Same here.

Simon: Thank you very much.

Ginette: A huge thank you to our guests for their time and participation today. Please let us know what you thought of this show and if you enjoyed the panel format by going to our website datacrunchcorp.com and we’ll see you next time.

Attributions

Music

“Loopster” Kevin MacLeod (incompetech.com)
Licensed under Creative Commons: By Attribution 3.0 License
http://creativecommons.org/licenses/by/3.0/