Evan: Hello and welcome to the Mining Your Own Business Podcast. I’m your host, Evan Wimpey, and today super excited to introduce Mary Stimson. Mary’s a big data analyst at Southwest Airlines. Maybe you’ve heard of them. I saw Mary, it was very privileged to see Mary give a talk at Data Science Connect Conference a few weeks ago talking about creativity, talking about her role.
Super exciting. We’ll get into that in a lot more today. Mary, thanks so much for coming on the podcast.
Mary: Well, thanks for having me. I’m excited to
Evan: be here. Okay, fantastic. Hey, to get started, can you give us a little bit just about your background and, and how you ended up in the role you’re in today?
Mary: Of course.
Sure. So I actually started in a very non-data sciencey background. I got my degree in accounting from Texas A&M. And then I worked in marketing for a little while and I’m actually getting a degree in theology now. So I really have professionally and formally before all this, no background in data science.
I came on to Southwest Airlines in a business analyst kind of role. Within the data governance world about four years ago. And within that we were doing a lot of metadata management, which was right in that in-between of data science, but really brought in my business interests. So I was able to kind of get involved there.
So we were doing metadata management with a tool called Collibra. You may be familiar with it. Sure. We were doing a lot of data scanning, getting all of our columns and tables, , information about them. And then on the business analyst side, I was working with stakeholders to try to find out what information, our data analysts might need from that actual metadata.
Then I moved into the data science role that I’m in now about. Two years ago, again, and I’m a big data analyst. That’s what we call them VDA now. So I started out a little bit more of a traditional data analyst role. For us, our big data analysts kind of step in on the front side of the data science lifecycle.
Okay. Doing a lot of investigation work, trying to. Take in what model we might need, understand all the features and factors that go into what the stakeholder needs, and then doing some early hypothesis testing, some early modeling to try to understand what will pass on to our, , more formal data scientists in order to build that more complex model.
About a year ago I kind of transitioned. I’m still a big data analyst, but I’m really more of a front end application developer for our models. So a lot of times, , our models spit out answers and then our users need to bring in their human intuition, which is much smarter, , to address what the model is suggesting, maybe override it, maybe put in some more factors that are useful, and then run it again and see what comes out of that.
So what I build is usually a web application for that kind of model where we can, , address that information and allow the user to really interact and use their human smarts to do what’s really important with models. And I always say that data science should provide information and not answers to people.
So I love what I get to do because it, it allows that intuition to, , interact with what computers are telling us.
Evan: Awesome. Yeah, that is, that is such an important role and I think. Oftentimes in the data science world, one of the most overlooked, you know, a a lot of focus on the, the backend and sort of what it seemed like your job earlier was preparing the data, making sure the right data is available, and then of course the, the modeling piece.
But then there’s the answer. Okay, go, go run with it. So real focus on user experience and, and I guess really the application development, the interface, sounds great. Can you talk. Who some of those users are at Southwest. I’m familiar with Southwest and as much as they’re an airline and they fly airplanes, but I suspect it’s not a pilot that you’re developing these applications for sure.
So can you talk about sort of who, who’s using this?
Mary: Sure. I specifically have developed models for finance and provided some interfaces for that group. , it really just depends on what we’re working on. My team is really fun because we’re essentially internal consultants, so lots of different departments might interact with us and take on projects, so we get to work on lots of things.
But right now we’re doing a lot of work for our operations group, which would mean for us operations is. How do we make bags move faster? How do we help a plane turn around more quickly between landing and take off? So it really might be any sort of leadership in those areas who can influence those factors.
Evan: Okay. Awesome. And so you, you mentioned your team, I assume it’s other big data analysts, maybe other people UX facing or can, can you talk a little bit about the composition of your team? Who, who’s on your team? Sure. What kind of skill sets do they need—do they have?
Mary: Sure. So our life cycle for our projects maybe is a good way to explain that.
We start with our leadership taking on projects, and then we have business consultants that, , work with our business stakeholders to try and kind of translate between the technical side of what we do and the business side of what our users need. And those business consultants kind of push us into projects.
Sometimes ask acting as project managers, sometimes ask acting as kind of mediators between the two and they’ll pass it on usually to the big data analysts, which we’ve talked about. And that gets passed on to our data scientists. Those are usually our more formal statistics kind of users, that kind of thing.
And then we pass it on to our business stakeholders. So that’s, that’s about who we’re made of. I’m kind of a unique little silo in that, being in the UI/UX space. I’m really the only one right now. But we also work with a sister team in it that those are our data science engineers. And a lot of what they do is helping us, , not only get our architecture set up correctly, that kind of thing, as, as our team sometimes is just setting up a model.
They help us stay on track with what we need to do and especially security wise, that kind of thing. And then they’ll help us deploy into whatever platform or environment we’re using, which right now a lot of times is AWS.
Evan: Okay. Awesome. Sounds good. And, and sort of from the, the business consultants through, through your team, the data science folks, Is there, you mentioned the term internal consultants, and when I think consultant, I think, you know, you have some business knowledge.
You, you, you understand operations at an airline, but not, maybe that’s not necessarily your formal background. Maybe you’re not the person with a degree there, and you’ve been 20 years. So on your team, is there a lot of, can, can you talk about sort of the level of depth of knowledge that’s required? For the business, like for the, the airline industry or even a specific function within the industry for operations within an airline.
Mary: Sure, sure. Across Southwest, it really depends, but in data science, it’s kind of interesting, like in the consulting field, you might be working on an oil project, but you’re certainly not an oil expert. Sure. And so for us, I don’t know very much about what a, a RAMP agent does or what a customer service agent does, but I’m excited to learn about it so that I can contribute to what they’re doing.
Within that. In our team, we have some people who’ve come over from other airlines and who have aviation backgrounds, but we also have a lot of people who come from very different backgrounds. I think you and I have spoken about this, that on my team, we kind of brag and joke that we have the highest ratio per capita of musicians across the company.
I don’t know. That’s great if that’s true, but we like to say it because we have a lot of people who come from all sorts of different backgrounds. I know a guy who studied jazz, a gentleman who was a professor in communications for the majority of his career, a gentleman with a background in history.
So we come from all sorts of different backgrounds and we have, of course PhDs in data science. But I would say within our team, It’s pretty diverse knowledge sets and, and more of it comes from the way you think than really necessarily your data science background and, and certainly more than your aviation background.
It’s, it’s something we’re all excited to learn, but don’t necessarily need to know before we come. Perfect.
Evan: Yeah, and I think it’s the way, it’s about the way you think, which is great. And Mary, are you a musician?
Evan: Okay. Just a little bit. Let me, let me just try to set you up to brag a little bit. Is there any way that’s specific about musicians that’s the way you think, some specific thinking or mindset that a musician would have that, that, that fits well in with the, with the team you’re on?
Mary: That’s a funny question, and it’s one that I’ve talked about with a few of my friends because it definitely does fit in in some ways, especially if you’re familiar with music theory. , there’s, you know, a lot of science between—behind how chords work together, how notes work together. , and a lot of that is actually physics, not necessarily physics that you have to understand.
Certainly not physics that I understand, but when you kind of are from that background, you end up thinking a lot about the logic of systems and how those systems work the same way that I would say my accounting background, even though I’m not really working on that, contributes because you’re thinking about how systems work together.
So I would certainly say from that sense, music is related, but also from the creative sense. I am, am just big on the idea that data science is a creative field, even though it might not seem like that to people that we know otherwise. And so I think the two do overlap that way. Okay.
Evan: Yeah, that’s great systems thinking in music. Mm-hmm. I was one of the probably very many folks who bought the first musical instrument at the beginning of COVID and thought, now’s my chance. Now I’m going to learn to play. And it’s, it’s gathering dust in the corner here now for the last three years. , love it.
But maybe I can at least check a musician box now. I own an instrument and so that helps. That’s right. Our data science team, I hope that’s right. Okay. I, I do want to give you a chance to talk about creativity maybe, maybe in a little more depth. I think music maybe is a good venue into that. Sure. But I watched you give a talk at Open Data Science Conference.
You’ve done some very creative work. Maybe can you just super high-level overview, talk about that work, and then maybe a little more generally, why is creativity important? Why should it be valued by, by a data science team?
Mary: Sure, sure. I would say, on the one hand, I would say that creativity is sometimes underestimated in data science because we think of data science as problem solving.
And some people will say, oh, sure, you know, that’s creative. Okay, but in reality, a lot of times data science is solving new problems. And when it comes to solving new problems, there has to be a willingness to create new solutions, new ideas. And so in that sense, I would say that data science is pretty creative.
And for me, of course, I enjoy the UI/UX side because there is, there are colors involved. It’s, it feels a little more creative. But I would say the field as a whole requires creative individuals, even just in the way that you think about, maybe it’s math, maybe it’s any, any sort of field can be creative in that sense.
For me, I think that came really to fruition during the pandemic, like you said, I think we were all trying to become a little different— a little more developed version of ourselves. And, and that was really before I had a heavy background in, in programming even and this is, you know, what, only three years ago, but I had a lot of time on my hands.
So the example that you’re probably thinking of is at the time there was kind of this emerging, silly idea, not silly, but that animals could be taught to communicate in a similar way that speech language pathologists might use to communicate with someone who is nonverbal. Less verbal.
Differently verbal. And a lot of times that’s a touch pad, right? You might have a board with buttons on it that an individual can press to say a different word, that kind of thing. So there was this kind of social media thing going on, and I had a new dog and I wanted to test it out, and so I started putting down a couple buttons and noticing that my dog was using them in context.
She was saying outside when she needed to use the bathroom. She was saying play when she was feeling playful, but that felt very anecdotal to me, so I decided to bite off way more than I could chew and built this big board. It turned out to be about eight feet long. I learned how to 3D print buttons, all these different things.
I learned some circuitry that I certainly did poorly so that I could track how my dog was pressing those buttons in context to try to. You know, I built a dashboard to try to see from an analytics perspective if I could actually trust what was coming out of the hypothesis that I had, which was, in fact, that it was not reliable, was kind of disproven by what I ended up seeing in those evaluations.
So for me, that was just kind of a silly, really fun project. Certainly helped me grow technically in my programming skills, that kind of thing. My understanding of statistics, contextual statistics on a very low level. But at the same time, I think it really helped me think about solving new problems with new ideas.
It certainly wasn’t a problem that I’d have to face before. Probably not a problem that a lot of people had had to face before when it came to animal communication devices. , kind of a silly field, but, , I think that really represents what, what my team does now, which is thinking about things that we haven’t had to think before and enjoying the process of developing upon those new needs.
Evan: That that’s great and I absolutely love that example. I think when, when folks want to think about what do data analysts, what do analytics, what are they capable of? It’s, it’s that type of problem that it’s, it’s maybe not even, here’s a problem, let’s try to solve it. It’s a new problem. It’s. How do we frame that?\
How do we think of what this problem is? How do we even define? And it’s not that there’s a problem sitting out there waiting to have analytics applied to it. Take some creativity just to find what that opportunity even is sure. Which is very cool. And I love that it’s, you’re able to take some of those lessons to, to industry, to the team that you’re on now.
And maybe that sort of begs the question. I think I always use that phrase wrong, but I use it nonetheless. Maybe that begs the question, how do new projects come to come to be there? Is there a creative force within your team that sort of says, this is an area we should explore?
Or is it more, you know, leadership has prioritized things. This is, this is what you guys should work on. Think of creative ways to, to try to solve that.
Mary: Definitely, I think it’s a mixture of both, which is also something that I really enjoy. We have projects that are passed down from leadership and a lot of times those come from different departments who are asking, Hey, we have this problem.
And, and maybe they just don’t have the resources. They don’t have the individual resources to solve it. So they might come to our team, can you set up a pod to help us work on this? But sometimes it comes from within our team. Hey, we see this need and we want to try to support it. We try to follow sort of an 80/20 rule.
So about 80% of our time should be on project work, but 20% of our time should be on personal development and growth technically, but also on innovative projects. So I think a lot of us have side projects going on that are just trying to solve a problem that we see a need for, or sometimes there’s not a problem.
It’s just fun. But so I think it would be both of those. And I, I’ve loved the 80/20 rule for, for personal support in our creativity.
Evan: Awesome. I I’ve never heard the 80/20 rule. The role that I’m in now, we, you know, we have. Something similar. We have some, some data hos where you work on some opportunities for professional development that is really just go, sort of flex your data creativity muscles.
I really love the 80/20 I, I like, maybe we can speak a little bit more about that. Are there, are there guardrails around the 20? Should it be, Hey, this needs to be focused on optimization because that’s what we’re looking at in operations right now, or it needs to be focused on this specific thing? Or is it more.
You’ve got a very wide canvas here and, and look for creativity
Mary: where you can. That’s an interesting question and it probably depends on the leadership for that group. I would say for me it’s pretty open-ended. If you have 80% of your capacity is on project work, the other 20% is, is free playtime as, as long as it’s within the scope of what we do.
I think it’s a pretty open-ended thing to. Like you said, flex those creativity muscles within our data work because otherwise we would miss lots of opportunities for important projects because if you’re just taking in what’s coming, that’s awesome. But if you really want to be a productive group, , pushing things outward as well is so useful.
Evan: Awesome. Yeah, that’s, especially when you’ve got a team full of creative musicians that can think, think about new things. One of the things that stood out to me in your role, you, your title, big Data Analyst, I think, you know, maybe, maybe a decade past Big Data was the, the term de yours. The, the, the pop culture referenced the term that was this is going to blow up, which now is maybe, I don’t know, maybe AI. But is there anything that differentiates big data or what does it mean to be a big data analyst?
Or I guess, is there even a data analyst at Southwest or is big data analysts just heads up? You’re, you’re coming here to a lot of data, a lot of fast-moving data.
Mary: Yes, I agree with you. Sometimes it’s a, it’s a term that’s certainly overused, but I think, yes, there are data analysts at Southwest. It might be a minimal differentiation.
I don’t know if that’s just a department to department thing. We’re probably all, some big data analysts in some ways. We’re certainly all data analysts in other ways. To me, big data, the term, the, the word that stands out right is big. It’s always been about volume. It’s always been about, you know, there’s more, what is it?
There’s, there’s more data coming out than we can ever understand coming in. And I think that’s really where it stands out to me that within an airline we are so big that there are always data points that are emerging that we have no idea are being taken in. And so within our team, I have peers who have been working here for 25 years.
They’re extremely familiar with our data landscape, but they will even never know the, the vast number of data points that we’re taking in there. Columns and tables that we have no idea exist until we just stumble upon them. And so I think that’s what big data means to me is that there’s always a frontier of unknowns.
There’s always a frontier of kind of that infinite growth that can be discovered. So I think that’s also what’s fun about. Big data is, is there’s always something to be, , excited about ’cause you don’t even know it’s coming. That’s what it means to me. I know it means a lot of different things to different people and, and it certainly can be overused, but in general I think it’s about an unknown frontier.
Evan: Yeah, I think that’s great and I think that that emphasizes the importance of creativity. It’s not, here’s your 20 columns. Go and build a model to predict this, this 21st one. Right? It is. Here’s an effectively unlimited amount of data. Which things are important to solve, which types of, of, of real problems that you have.
So yeah, I think it goes, goes hand in hand with, with the creativity there. I’ll, and maybe you’ve already sort of covered this, but the. The other term du jour for today, if of course is ai, AI is everywhere. Yes, you mentioned sort of the, the connection your team has with the, the data science and the modeling team.
Is, is AI something that, that they undertake? Are they, they using advanced models? Is AI more than just a term that goes on a PowerPoint bullet point? Is it, is it something that actually gets used?
Mary: Sure. Yes. It definitely is. Especially when we. Compare AI and ML machine learning. Certainly both of those are important to us at Southwest.
I think speaking at a high level, I, I think it is a, a frontier for data science in general. It’s something that we’ve been using for a while, but also want to learn a lot about how we can continue to in, involve that, in what we’re doing going forward. I think of—not to sound too corporate here—but our CEO Bob Jordan has five objectives for us in the next, , couple years, and three of them really stand out to me as I think about this in terms of where AI can step in.
And these are just kind of a, you know, Ideas that I have that I think would be interesting, but one of those is to modernize the opera, the operation, excuse me, I’m thinking about operationalization of the model, right? Modern, modernize the operation, which for us means, , you know, how quickly can we, like I mentioned, take from the moment the plane lands to the moment it takes off.
There’s so many features in there and sometimes that’s something that we’ve been looking at for so long that I think AI can step in and, and use a different perspective to help us understand. Sure. , we are trying to maintain our low cost advantage, and so for us, you know, Southwest has a lot of deals that make us low cost, but at the same time, we still fly aircraft.
It’s still expensive to fly an aircraft, so how can we price the market well and understand supply and demand in, in a way that does help us understand what our human intuition is telling us now? And sometimes just what our models are telling us now can be limited. The last one that I think of, one that I would love to work on, I think it would be so much fun, is, is that our, one of our values is to be a good citizen, right?
To try and. Do good things with our aircraft to connect people to what matters in their lives, but also just to be involved in the world and in a helpful and wise way. And, and one of those goals that we have is to reduce our carbon emissions. And I think that is so cool and so much fun and there’s so much opportunity for AI and ML there to step in and help us understand small areas, big major areas where we can reduce those things.
Just because it’s such a complicated area. So I think one project I would love to work on would be to see how we can use AI and ML to reduce our carbon emissions. Man, wouldn’t that be fun?
Evan: Oh yeah—okay. You’ve got 20% of your time. You can, you can devote that.
Mary: That’s right. Oh, I do not know enough about that to, to even pretend.
Evan: Well, actually may, maybe you’ve just answered it and I’ll offer a, a, a chance to give an alternative answer as well, but especially in, in your shoes where you’ve got sort of the flexibility to, to pursue these, these creative endeavors, you know, within the scope of, of your field and the work that you do.
Let’s, let’s say, you know, you just wrapped up all your projects and now your 80% is gone too. Mary, you’ve got a hundred percent of your time to. To find a new thing, and we’re giving you as, as, as much breadth as you can, and you’ve got the team around you too. So all of the infrastructure team, the IT team, the data science team, the business consultant, they’re all aligned to whatever your vision is and the thing that you want to pursue.
So where, where do you go there, where, where would you spend your time in this sort of dream scenario?
Mary: Oh, that’s a fun question. Yeah, that’s a fun question. And. A few come to mind. The first one would be carbon emissions. Gosh, that would be neat. The second would be if we ever needed an animal communication device on an aircraft, I’ll certainly step in there.
We’ve already done it. It’s already built. Yeah. We definitely don’t need that. The third that comes to mind that I think Southwest has been a fore runner in and that I would love to participate in, especially from a data science perspective, would be how do we support our travelers who have disabilities?
There is so much conversation around how we can really make that a, a positive experience for someone who maybe can’t navigate an, navigate an airport the way that I can or get it on aircraft the way that I can. And so how do we make that a really positive and encouraging experience for someone who, , might be trying to navigate a space that’s not comfortable for them.
And I think as much as that might sound, sound like more of a qualitative challenge, there’s a lot of quantitative opportunities in that when it comes to mapping an airport. We don’t own our airport spaces. We lease them, but we do have a lot of influence over how they’re used. I.
So I think when it comes to how are our facilities being used, how is, how are Southwest facilities being navigated?
Even just when it comes to, you could do heat mapping in an airport, right? To see which areas are most traveled, which areas do people need to access the most, where we might want to help someone access, but also which areas are a little bit less accessed, that we could help someone who needs help navigating operate in.
So I don’t know if that makes sense, but that certainly would be an area of interest of mine because I know that it’s something that Southwest is passionate about and definitely something that, that I think would be really interesting to bring in data science in an area that usually would be considered a little more qualitative.
Evan: Awesome. Yeah, that sounds like a great answer. And it also sounds like it ties to your. Unique role of thinking about these problems and how to apply them, and also thinking about the user experience and the user interface. This is more away from the application and into the real world, but it’s still a focus on the user experience, which is great.
It’s been super interesting, Mary, to talk to you today. That that is the last question. That is all the time I have here. Is there anything folks should know about you? Should they follow you? You’re speaking at another conference? At any point that we should point folks to?
Mary: Sure. I love the Data Science Connect conference. I’ve gotten to speak there a few times. That’s where you and I connected, Evan, and so I would really encourage people to try and visit and be involved in that if they can. They have virtual opportunities. There’s also one in person in October. I don’t know if I’ll get to see you there.
That would be fun. but I would really encourage that people might consider that conference. It’s a great resource. It’s a great opportunity.
Evan: Okay. Awesome. Mary, thanks so much. If you are going to attend in person, try to fly Southwest. Report any bugs in the system to Mary Stimson and she’ll get them fixed for you.
Mary, thanks so much for coming on the show today. Super enlightening conversation. Really, really happy to have you.
Mary: Thanks, Evan. It’s been a pleasure.