Data Science Mixer

Tune in for data science and cocktails.
MaddieJ
Alteryx Community Team
Alteryx Community Team

Samsung’s Charity Wilson joins us to share how her early aptitude for statistics has led her to a rewarding career in HR analytics, and how causal inference strategies fuel her work. 

 

 


Panelists

 


Topics

 

 


Cocktail Conversation

 

Share a skill that you learned in school and never thought you'd use, but it's turned out to be valuable in your career!

 

Mixer LI (1).png

 

Join the conversation by commenting below!

 


Transcript

 

Episode Transcription

SUSAN: 00:00

When did your interest in data begin? College, high school, or maybe even earlier? For Charity Wilson of Samsung, today's guest, it was high school. More importantly, it was discovering that she could have a positive impact on other's lives through statistics and data analysis. That's something she's still passionate about today in her work in HR analytics, her exploration of causal inference strategies and her projects using data for social good. Welcome to Data Science Mixer, podcast featuring top experts in lively and informative conversations that will change the way you do data science. I'm Susan Currie Sivek, senior data science journalist for the Alteryx Community. Let's meet Charity and get the details on her inspiring and innovative work.

CHARITY: 00:52

My name is Charity Wilson. I use she/her pronouns. I work for Samsung within the HR department. I work specifically on HR analytics.

SUSAN: 01:03

Okay. Awesome. So yeah, that's really an interesting area and kind of a hot area right now as well. Tell us a little bit about how you got into your data career. What was your journey toward that path?

CHARITY: 01:14

My journey was probably long and idiosyncratic. I grew up out of data science in high school, to be honest. I was recruited by the PE teacher to do statistics for our high school basketball team. So that was my first point where I learned that data analytics had the power to make a difference. Fast forward, I was a single mom trying to figure out how I could supply my kids and meet all of their needs, so I pursued an MBA, not knowing exactly what it was that I was going to do with an MBA. But in the first class of that MBA, it was economics, and the professor talked about econometrics is a growing field that allows you to utilize data to figure out which data points are the most important and which ones answer business questions. And he lit a fire underneath me. When I finished my MBA, I went straight on and got a second masters in econometrics. At the same time, I've been pursuing careers in the data science world. I worked for Concentra, which is in the medical arena, understanding customers and patients, what things about their satisfaction and what things about their visits impacted them, just understanding what it was that really moved them to say a certain medical visit was good or was bad.

CHARITY: 02:52

Then from there, I moved into consulting. I worked with Saxony Partners and worked in a variety of fields. Most of their stuff was focused on banking and real estate, which happened to coincide with my dissertation for econometrics. I did an in-detail analysis of real estate, specifically at the affordable housing crisis, really looking at how businesses could meet the needs of people who needed affordable housing and make it profitable at the same time. Right now, I'm working for Samsung within HR, so a different pivot. Looking at the life cycle of employees, what are the things that matter? How do people make decisions about promotions, about attrition, about hiring? What are the things that push and pull people?

SUSAN: 03:48

So I want to go back to something that you mentioned just in passing there that I think is really neat. So you mentioned that you started actually doing statistics in high school. I'm not sure that many of our data folks can trace their data lineage back quite that far. Is that something that you actually at the time thought was fun and interesting, or was it kind of like, "Yeah, okay. I'll do the stats"? Or did it really kind of spark your fancy at that point?

CHARITY: 04:11

It really did spark my fancy. I was a geek in high school. Let's just be totally honest.

SUSAN: 04:17

Well, I think you're the only data person who could fit that. Yes. That's very rare.

CHARITY: 04:24

Yeah. My nose was buried in every single science fiction book, but I was jealous of all of my classmates who played sports and were looked up to. So when I discovered that I could take my math skills and really make a difference in something that was popular in high school, it was a really powerful moment. It was funny, at a class reunion, one of my classmates, he looked at me and he goes, "I will always remember you at basketball." Because once I started with the girls' team, I started doing the boys team too. His team was playing and he went up for a free-throw shot. And as he was going up for the free-throw shot, I mutter to myself, "Oh, crap." And I muttered his statistics for making free-throws shots. At this point in the game, I had it memorized. I already knew he was going to land the first shot and miss the second shot, and I was just [screwed?]. I was infuriated. What is he doing in at this point? He should have already been pulled from the game. I know that he's already reached his fatigue curve and he needs to be pulled from the game. He's going to miss this second shot. And his parents happened to be seated right behind me. Like, "Oh, dear."

CHARITY: 05:48

So it was in high school that I discovered the power of statistics to help these coaches understand the fatigue curve of high school players. This is the point at which this player needs to be pulled from the game so that they can rest and then they can play in the fourth quarter. This is the shooting percentage that we see when they're reaching their fatigue curve. So it was a powerful moment for me as a geeky high school student. I didn't think that I had a lot to offer. And in those moments, I discovered that I really did have a lot to offer.

SUSAN: 06:24

Yeah. Well, that's great. Great story. And I would imagine that there's something of a straight line too between looking at players and coaching strategies and so forth and people analytics today in the kinds of things you work on now? Is that a fair guess?

CHARITY: 06:40

It is. It is. Yeah.

SUSAN: 06:41

Awesome. So tell us about some of the projects that you've worked on, as many technical details as you would like to share or feel comfortable sharing or able to share. Maybe pick a favorite project or two and then walk us through it.

CHARITY: 06:54

Certainly. A favorite current project is currently I'm working on things related to causal inference. So what I'm working on is what is the influence of a particular bonus on someone's decision to quit their decision to continue with the company? Because there are a lot of bonuses that are offered, that are utilized. And it's incredibly messy data because all these things are correlated. It's really hard for our HR leadership to see what is the ROI? Should we pay higher bonuses? Is it going to be worth our while? Should we pay lower bonuses? Would that get a better return? What's going to achieve the best outcomes for the company? So these are really curious and wonderful projects that have the potential to really provide good outcomes to the employees.

SUSAN: 08:06

Yeah. Definitely. And I can imagine that trying to separate something like a bonus from all of the other factors that are included in somebody's compensation and work life. I mean, isolating that as a potential cause for retention must be pretty tricky.

CHARITY: 08:24

It is. A lot of analytics are correlated. You look at things that run together at the same time. So a lot of the bonuses are correlated with a lot of other things, just as you would expect when you have an employee who is performing extremely well. You're going to see things like really good review scores, various lateral moves, promotions, raises. So teasing out the individual impact of one variable is incredibly hard. It's just like looking at sabermetrics. What is the impact of one single player on the team's ability to win a game or get more runs? It's because there's such a team impact of everything, it becomes really challenging to say what the impact of one single thing is. I was doing one analysis for one smaller division, and unfortunately, I ran into the fact that promotions and raises were highly correlated. Kind of intuitive.

SUSAN: 09:37

Sure. Yeah. Makes sense.

CHARITY: 09:39

So you get to the question, what's the impact of the raise on retention, on somebody's willingness to stay with the company? Well, because they're so tied together, it becomes really, really hard to pull it apart and say, "Oh, this is the impact of the promotion." Unfortunately, within econometrics, one of the key principles is independence of variables. And in this case, they are not independently occurring.

SUSAN: 10:13

Right, right. So can you tell us any details on what techniques you're using to look at those variables?

CHARITY: 10:20

Certainly. Yeah, I can't go into a little bit of detail here. I use a type of causal inference. It's funny. One of my profs, we covered a whole semester on causal inference. And this was a one-hour lecture that he described how to do this. And he stated, "You probably won't ever do this in real life. It requires a huge data set."

SUSAN: 10:46

Oh, wow.

CHARITY: 10:47

And it's unlikely you're actually going to ever do this.

SUSAN: 10:52

And then the time came.

CHARITY: 10:54

Then the time came. So what I do is it's essentially a double-blind experiment done in reverse, so kind of like COVID-19 vaccine tests. They did this in the right order though, and I have to do it in reverse. So they gave the vaccine to a randomly assigned group of people and then waited to see what is the impact. Do we have people get sick? Do we have people survive better? What happens? Well, unfortunately, bonuses are not randomly assigned. And I don't think I could go and tell HR leadership, "Hey, I'd like to run an experiment. Let's randomly assign the bonuses. How exciting." I don't think I would get by it.

SUSAN: 11:43

Probably not.

CHARITY: 11:45

So what I do instead is I take the group that got bonuses and then I use a matching algorithm, specifically, Nearest Neighbor. And I go through and I say, "Match the people who got the bonus with the closest individual who did not get the bonus." And then I take those matched pairs and I run them through a regression analysis. So the only difference that should be between these two groups is the bonus. And if that is statistically significant, then we know that has a causal impact. So sometimes though, getting them to match, I can't. I discovered promotions were predominant in the group that got the bonuses and they weren't as predominant in the group that didn't. So it's a really tricky, finicky algorithm that fails more times than it works. But when it works, I can go back to HR leaders and go, "Yeah. This bonus has an impact. This one will retain your people. It will prevent attrition. It really does work."

SUSAN: 13:12

Nice. Yeah. And that's a pretty powerful statement to be able to say, "This really does work." So I'm sure they're excited to hear something more definitive like that.

CHARITY: 13:20

Yes. Yeah. They are. They are. Yeah.

SUSAN: 13:22

Cool. Is there another project that you wanted to share about?

CHARITY: 13:25

Certainly. I did a lot of research into the affordable housing crisis here in the Dallas Fort Worth area. This was research conducted back in 2017, 2018, and a little bit into 2019. So it didn't get into the pandemic and how that's affecting housing. We all know that the housing market has been very volatile, given the pandemic and all of the different things going on. So what I was looking at was the price scene on specifically apartments. Apartments are referred to as a multifamily dwellings. But I also did look a little bit at single-family housing. And the thing that kept on coming up was the price per square foot as you go up different qualities. Most of the time, the price per square foot has a nice stair-step increase as you go up and quality. The better apartment, you're going to pay a lot more for. But there were some points where it didn't. The big one was-- most people don't realize this. Apartments are graded. You can get an A-plus, you can get a B-minus, you can get a C-plus apartment, and they mean exactly what you would expect. The real point of differentiation was as you stepped from the B-plus down to the B and the B-minus, it didn't step.

CHARITY: 15:14

Essentially what it said was, if you happen to own a multi-family housing dwelling, don't invest in upgrading the property all the way to a B-plus because the demand for a B and a B-minus is so high, you will never recoup that value because, as we said, the affordable housing crisis was huge. So what this suggested to all of the people who were investing in multifamily housing was instead of going after your luxury properties and getting those ones that give you the nice big dollar amount, you can get almost a similar return if you go and invest in your B and your B-minuses and you keep them at that level because that's where the demand is. And you're going to make a better return on those than you could make on a B-plus or an A-minus. If you meet that demand for those affordable housing, the demand is so heavy, you're going to get a good return. So for Pete's sake, quit going after your luxury apartments and quit taking those apartments that are meeting the needs of your working-class people and pricing them out. Keep them instead available for those working-class people, because you're going to give us really good, steady return.

CHARITY: 16:50

One of the other things that was really interesting, there was a response to unemployment, and it took me a while to figure out what it was. There are certain fields in which people work that are not as elastic. For instance, somebody who works in car repairs, when there's a downturn in the economy, they aren't so likely to lose their job, until we get to something like the pandemic where people aren't driving as much. But there were these fields that lacked the same elasticity, and when you went after those, let's call them working-class people, you could get people who were more robust and could respond better to unemployment, to dips in the economy, and could keep that apartment complex filled with lower rates of people having to move out. What I realized was don't downplay the ability of the working-class people to be incredibly profitable and more robust and provide a better return on investment than the people who occupy your luxury apartments.

SUSAN: 18:14

Right, right. Yeah. It's interesting. When you first started describing that result, I thought, "Oh, does this mean that landlords should not be investing money in apartments to fix them up? Right? Which on the one hand sounds kind of negative, but it sounds like that is actually the case, that if the goal is to keep people housed, then the better strategy is to have apartments that are less fancy, but still affordable at that level.

CHARITY: 18:41

Correct. I did do research into a little bit about the repairs, and I did have to control for landlords who were making the decision to invest in things that needed to be done to keep the ACs up to date, to work on the roofs. So there was a level of investment into the property. It's just that it wasn't the excessive, huge investment that would raise a property up a grade. Instead, it was what was needed to maintain the apartments at the level that they were and continue to serve the demographic that they were originally serving. But they did have to. It was essential that a certain level of maintenance was being done within these to maintain the current clientele.

SUSAN: 19:38

Right, right. So no infinity pools, but definitely keeping the roof intact and all that good basic stuff. That makes sense.

CHARITY: 19:45

And keeping the basic swimming pool in the same condition that it was when they got ownership of the complex.

SUSAN: 19:52

Definitely. Definitely. Cool. Was this research that you were doing independently, just a topic that you were interested in or for some other organization?

CHARITY: 20:01

While it was at Saxony Partners, Saxony Partners happened to have a software that was utilized a lot within multifamily housing dwelling. So there was a lot of data that they had available about multifamily housing. So as I was approaching my dissertation within econometrics, I talked extensively with my manager and talked extensively with the CEO of the company, and we made a proposal together to do this research. What sort of insights could we gain out of the data that we had in-house? And could we move this data into more of a consultative role so that we could advise our clients who owned these multifamily dwellings, advise them on things like which properties to move ahead and purchase, which ones to invest heavily in an upgrade from a B to a B-plus? So this was a project that was both school and work and my own curiosity about the market. So it was a lot about what the clients asked and wanted, what they dreamed up, so that we could answer questions they hadn't even started to ask yet.

SUSAN: 21:33

Nice. So you're fulfilling their data dreams, but also getting your school and work projects done. I love it. Multitasking at its best.

CHARITY: 21:41

It was fun.

SUSAN: 21:42

Awesome. So let's turn now back to some of the things that you are working on these days, focusing mostly on people analytics and data science for HR. What are some of the issues that you've encountered in that area for folks who haven't done any data science work in that field? Are there some unique issues that you think come up with that specialty, and how do you deal with that?

CHARITY: 22:05

One of the issues that I deal with is clustering. Human beings tend to group, which is why we see things like Little China and Little Sicily. People go places where they see other people that they think are like them, where they see people and they can identify with. So as a result, within education, certain demographics are more likely to have degrees in electrical engineering and other demographics are more likely to have business analytics. And then there's the weirdos like me with the econometrics degrees. So people group and econometrics style algorithms expect that the data is randomly distributed, and it's not. There is not random distribution going on. So teasing out the effect of-- for instance, one thing that's highly correlated: age and educational attainment. I'm sorry. I have yet to see an 18-year-old who has a PhD. I know that they exist.

SUSAN: 23:20

Definitely the outliers. Yeah.

CHARITY: 23:21

There are a few of them. Yes. They're outliers. But age and educational attainment are heavily correlated. So teasing those two out, when am I dealing with the effect of age? When am I dealing with the effect of educational attainment? This morning, I was running an algorithm, and I'm looking at attrition, and there's an attrition curve that goes over a lifetime, but it also goes over educational attainment and which one is driving the train here. And just figuring it out, all of my regression analysis, I also put in a correlation metrics because I have to check all of those variables and make decisions about, "Oh, this is too highly correlated. Okay." And most often, age and education are more highly correlated. And in this data set, which one of them is driving the train? And then I'll keep that one and drop the other.

SUSAN: 24:41

Right, right. And I imagine that's, again, where those causal inference approaches come in handy when it comes down to it.

CHARITY: 24:48

Yes.

SUSAN: 24:49

Awesome. So going to try to bring this full circle a little bit. Are you still a sports [inaudible]? And you mentioned sabermetrics earlier. I'm wondering if any of your previous sports experience and maybe current fandom possibly come into play with your people analytics work.

CHARITY: 25:07

It's funny when people hear that the start of my day, a data science journey began with basketball analytics. I oftentimes get asked, "Oh, can you predict who's going to be in March Madness? Can you predict that?" And the truth is, I haven't specialized in that area. And I do study sabermetrics and I do look at the things that are coming out of team analytics. They are so helpful in understanding feedback effects, but no, I don't follow sports. I follow gymnastics, especially right now. But no, I don't actually follow sports. I do follow sports analytics because it really does push the field in different ways.

SUSAN: 26:11

Interesting. So tell me a little more about that. You mentioned the feedback effect and how that's been something that has been informative and relevant for you.

CHARITY: 26:20

Correct.

SUSAN: 26:20

And tell us what that means in the first place, because I'm not entirely sure I know.

CHARITY: 26:25

Let me go back to my high school experience, and then I'll fast forward. The first year that I sat on the basketball team, we happened to have a pair of identical twins who played on the team, Letta and Marret, and they were phenomenal basketball players. The two of them were just amazing. There was an effect though when the two of them were on the court together at the same time. Everybody's metrics improved, rebounds, shooting percentages. There was an effect. Because these identical twins were playing and had this really good synergy between the two of them, it improved the synergy of the whole team.

SUSAN: 27:17

That's so interesting.

CHARITY: 27:18

So that was actually something that came up recently in some of my research. One of my HR leaders asked, "Hey, can we look back and see what bonus strategies have been really effective?" And I kind of sat there. What she wanted to know is "Give me a recipe of putting together a team of bonuses that are more likely to result in people being retained." And I just went, "That is not a question you can answer with regression." Oh, backward. How in the world do I even begin to answer this? And it was like the effect of having the twins on the team. What were the teams that were more likely to result in retention? What were the group [sovidence?]? And what came to mind was a session at Inspire on Market Basket Analytics. And then I was looking out in the community and I happened to find your article on Market Basket Analytics.

SUSAN: 28:32

Awesome. Yay.

CHARITY: 28:32

And I went, "Oh, my word. Oh, my word. This is how we do it. It's a basket. It's a basket of bonuses. Which basket of bonuses results in retention?" And it was just, "Light bulb."

SUSAN: 28:54

That's a great moment.

CHARITY: 28:55

And oh, my word, the HR leaders, they love it. They absolutely love it. They're like, "Okay. So this bonus plus this bonus plus a promotion." And then I've got it filtered down to we can now see what are the twins that you'd put on that team and suddenly you've got all the synergy that you need to have, that feedback between a, narrowing it down to these are the twins. The other bonuses that are correlated with them, they're correlated. They're great. But these are the twins that you put on that team and your retention or your attrition rates show a significant improvement, decrease whatever it is you need.

SUSAN: 29:49

Awesome. Wonder Twins is one thing that comes to mind. I have to admit though, my analysis of the situation when you first mentioned the twins was much less sophisticated. I just thought, "Oh, they were confusing the other team and that's why everybody got better." But no, much more a data-driven, nuanced perspectives, so.

CHARITY: 30:05

Well, Marret were wore her hair short and Letta had long hair, so it took some staring at them to realize that they were actually identical twins.

SUSAN: 30:16

Oh, that's so funny. Very cool. I love though how these little experiences that we've had in the past sometimes come up and can help spark those light bulb moments. That's awesome. So I know you've also been involved in some other kinds of projects where you've used your data skills for social good, other kinds of projects. Would you like to tell us about one of those and that experience?

CHARITY: 30:36

I worked with the Commit Partnership, which is the recipient of an Alteryx for Good grant here in the Dallas Fort Worth area. The Commit Partnership aggregates data related to schools and kids and performs analysis on it and makes this data available to schools, universities, anybody that has the right and privilege to be able to see the data. Data on school quality is hard to find, and they really meet that need and do a lot more than just meet that need. What I did with them was I really started digging into what's the effect of stated gender and poverty on school scores and kids scores for standardized testing here in Texas. And I was able to quantify the impact of when a kid is on free lunch program, which is a measure of poverty. I was able to quantify that impact and also quantify the impact stated gender on reading scores, math scores at different grade levels. One of the things that we saw was in the younger grades, gender has very limited impact on the kids' scores. But once you get past fifth grade, sixth grade, then it starts to be statistically significant. Girls' scores are higher for reading. Boys scores in math are higher. And there really is an impact. It's a socialized impact, but it's real. The impact of poverty though is just as much as the impact of gender.

CHARITY: 32:26

I absolutely loved working with the Commit Partnership. It was really powerful. There was a moment when there was a bill going before the Texas House and one of my liaisons at the Commit Partnership called me up, and he's like, "Okay. Look at this and give me a what-if analysis." And I had all the data. I had all the algorithms, and I ran it through and I told him, "If we can get that money, it has to be this amount to make a difference." He's like, "Oh, they're not even talking about that large." He goes, "Okay. We're going to pivot." And so they pivoted to-- I forget what it was that they put their emphasis on, but my analysis allowed them to pursue something that would be more effective rather than asking for something that wouldn't have been enough to make a difference.

SUSAN: 33:30

Nice. That has to be a good feeling to know that you're affecting their legislative strategy and helping cause some positive change there.

CHARITY: 33:37

It was truly powerful. It was one of those moments where I could say I'm making a difference for people, for real people. I grew up with parents who are missionaries. I grew up with parents who want to change the world, and I want to be my parents' daughter. I want to go out and make a difference for other human beings. So that was one of those moments where I knew beyond a shadow of a doubt that I was making a difference for kids here in Texas. I knew that decisions were being made that were effective, that would truly improve educational attainment.

SUSAN: 34:25

That is a really cool story. It's wonderful to know that one can have that kind of impact and truly use your data skills for good in that way. It's awesome to be able to make a difference in that area, which is so needed. So good for you. That's super cool.

CHARITY: 34:38

Thank you. Thank you.

SUSAN: 34:40

So we do have a question that we always ask on the podcast, and this is our little alternative hypothesis recurring segment. So I'll ask you the same question that we ask everyone who comes on the show. What is something that people often think is true about data science or about being a data scientist, but that you have found in your experience to be incorrect?

CHARITY: 35:02

For me, I find that people think data science is one huge field and I happen to have all of the skills of data science. People often don't realize that there are very specialized fields within data science. I was listening to the podcast. I forget his name, but the gentleman from South Africa who specializes in LP analysis.

SUSAN: 35:29

Yeah. [inaudible].

CHARITY: 35:32

Thank you. I totally respect what he's doing. My mother has her PhD in linguistics, so that was fascinating to listen to, and I have absolutely no clue how to do any of that.

SUSAN: 35:49

I love it.

CHARITY: 35:50

I just--

SUSAN: 35:51

Yeah. It's very specialized. Yeah, yeah.

CHARITY: 35:54

I analyze numbers. I don't analyze words. One thing I run into quite often is I'm not a visualization expert. I've always worked in teams where I've got someone who's a visualization expert. Recently, we were working on this analysis and I talked to one of my teammates. I'm like, "Hey, can you do it like this?" Because what we wanted to show was this is what's forecasted. This is the confidence interval around it, and here are the ones that are falling outside the confidence interval that maybe we should look into for manual review. And so I was describing it to her, and she put it together, and it was amazing. I can't think of those things. I can visualize in my mind a quadratic a relationship between variables. And when I see it in Excel, I'm like, "Oh, look, we've got a turning point there and a turning point there." And that means that we've got a law of diminishing returns within this state. I can do that, but I can't think how to take calculus and put it towards incredibly smart people who will understand it if I make it clear. No, no. I think of very complex crafts.

CHARITY: 37:16

So for me, data science is incredibly specialized and a lot of us have very specific skills, like my teammate, who can do fantastic visualizations, like that gentleman who can do amazing in LP analysis. But I have incredibly specific skills. I probably cover a wide range with those skills, but my skills are within the tool belt of econometrics. I take econometrics models and apply them to the business world and use that to provide insights to companies. Yeah.

SUSAN: 38:06

Yeah. That's awesome. What is funny, it does kind of bring us all the way back to this idea of forming the perfect team, right, and finding those perfect sets of twins and people with the ideal matched skill sets so that you can get that work done of using the data to produce those business outcomes. So yeah, everybody's got to have their specialty, but we put them all together and that's where the magic happens.

CHARITY: 38:29

Right. And that is kind of a project that we would like to work on. We don't have the data right now, but understanding we've got certain divisions within Samsung where we know if we lose one individual, there is such a team dynamic, that it can take over a year and sometimes never for that team to recover and start producing at the same level that it previously was. So you're absolutely right about that team dynamic and understanding-- we would love to understand, can we lose an individual and continue producing? Is there another individual that we could transfer in and keep the production level at what it is? But sometimes these research teams that are pushing digital and technology development forward, sometimes they really are that. It's this team, and that's it.

SUSAN: 39:29

Right, right. No, it's definitely challenging. And if you don't want to speak to this, it's totally okay. What kinds of data are you hoping to get to answer that question?

CHARITY: 39:40

We really need to be collecting data on skills and experience, and unfortunately, that isn't data that we currently have within our database. We don't have a systematic way of understanding what are the skills, what are the experience of the various team members, what is it that makes them good at their job? Right now, we've got a very transactional database that includes stuff like bonuses, promotions, education, but it doesn't include that detailed technical information about these are the skills that the person has. We've got somebody who can code and SQL. We've got somebody who can code in Python. They've got four-plus years experience doing that. And we don't have that in a systematic way so that we can understand how these teams are functioning the way that they're functioning, what and whom we could sub in. Going back to basketball, who can we sub in? field? Who is our point guard? Unfortunately, we don't have that level of data so that we can understand what our substitution options are.

SUSAN: 41:00

Yeah. That's really interesting. And then, of course, there will be kind of those inevitable qualities of the people on the team that are really hard to measure too even beyond that, so. Interesting, challenging problems.

CHARITY: 41:12

Yes.

SUSAN: 41:13

Keeps things lively, right?

CHARITY: 41:15

Yes, it does.

SUSAN: 41:16

Awesome. Well, Charity, thank you so much for telling us all those great stories and sharing all about your projects. We really appreciated it.

SUSAN: 41:24

Thanks for listening to our data science mixer chat with Charity Wilson. Joined us on the Alteryx Community for this week's cocktail conversation to share your thoughts. Let's keep it light this week. Charity mentioned that she first learned about a causal inference technique in a class where the professor said she'd probably never use that approach in real life. Do you have a similar data-related concept or skill that you learned in school and never thought you'd use, but it's turned out to be valuable in your work? Tell us about it. Share your thoughts and ideas by leaving a comment directly on the episode page at community.alteryx.com/podcast or post on social media with the hashtag Data Science Mixer and tag Alteryx. Cheers.

 


 

This episode of Data Science Mixer was produced by Susan Currie Sivek (@SusanCS) and Maddie Johannsen (@MaddieJ).
Special thanks to Ian Stonehouse for the theme music track, and @TaraM  for our album artwork.