Capital club radio

Download 63.18 Kb.
Size63.18 Kb.


Interview with founder Jennifer Priestley

00:05 Intro/Outro Speaker: Broadcasting live from the Pro Business Channel studios in Atlanta, Georgia, it's time for Capital Club Radio, brought to you by FLOCK Specialty Finance. Please welcome your host, Chairman and CEO, Michael Flock.
00:22 Michael Flock: Good morning, and welcome to Capital Club Radio. We're delighted and honored today to have a good friend of mine, actually, and a leader in the consumer credit analytics segment with us, Dr. Jennifer Priestly, PhD, Professor of Applied Statistics and Data Science at the Kennesaw State University, where she's Director of the Center for Statistics and Analytical Services. Jennifer oversees the PhD program in Advanced Analytics and Data Science, and teaches courses in Applied Statistics at the undergraduate, Master's and PhD levels. Prior to receiving a PhD in Statistics, she worked in the financial services industry for 11 years, including positions at Visa, Europe and London, MasterCard, as well as AT&T Universal Card and Andersen Consulting. She earned an MBA from Penn State and a BS from Georgia Tech. Jennifer, welcome and thank you so much for coming. You've got such a great life story, I hope someday I can write your biography; it's fascinating. But let's start at the beginning. Tell me a little bit about yourself, and as a young person, kind of what were your dreams and did you ever, as a young person, see yourself teaching Applied Statistics and Data Science?
01:36 Dr. Jennifer Priestly: Sure. So I think I can say with 100% confidence that, as a young child, I did not envision myself being a professor of Statistics. To be honest, I don't know that that really manifested itself until fairly late in my career. You had highlighted a little bit about my background prior to coming into academia, I did work for 11 years in the private sector and now I've been in academia for 11 years. I should probably highlight the fact that I started working when I was six.
02:07 MF: Six?
02:07 DP: No, just kidding.
02:10 MF: What, cleaning up your room or...
02:11 DP: Yeah, exactly.
02:13 DP: It just makes me sound very old.
02:14 MF: Okay.
02:15 DP: No, but I always did... When I was much younger, I always did think that at some point I would find myself in the teaching profession in some capacity, I just didn't know what form that would take. So I'm honored now to refer to myself as a teacher.
02:29 MF: Okay. And... But what's interesting, too, is you started out, though, in business, in the credit world with Visa, MasterCard. How was the transition going from business to academia? And why did you do that? Did you not like the corporate world or...
02:48 DP: Right. No, I actually loved what I was doing, to be honest. I did have about a million frequent flyer miles before I was 30.
02:57 MF: A million before you were 30?
02:58 DP: I did.
03:00 MF: How does that happen?
03:00 DP: Well, so after I finished my MBA, I went to work for AT&T and transitioned to the credit card side of AT&T. Eventually went to work for MasterCard in New York and had to travel around the country, working with different banks, as they were part of the MasterCard portfolio. And then eventually found myself at Accenture, and did an immense amount of travel when I was at Accenture, flying all over the place going to clients. And then, obviously, eventually went to work for Visa in Europe. Visa, obviously being based in San Francisco, I made a lot of trips from London to San Francisco and back, and most of my clients were either in Scotland or Ireland or Belgium, and so did a lot of travel at that stage of my career. And ultimately, my husband and I just got to a point where we thought if we were going to eventually transition to having children and having a family, it just didn't seem to be the right lifestyle for that. So that ultimately was the catalyst to transition to doing something different, and so I was fortunate to be married to somebody who said, "Sure, we want to lose half of our household income, let's make it happen." And so, I retired and I pulled out my blue jeans and my backpack at the age of 32 and went back and got my PhD.
04:15 MF: Awesome.
04:16 DP: Yeah.
04:17 MF: Awesome. So how was the transition? Because it's so different, being in a giant company like Visa or MasterCard, and in business, going to a world of professors and term papers and PhDs and all that good stuff.
04:34 DP: Yeah. Sure. So I think the biggest sort of epiphany or learning that I've had is that you can't substitute practical experience for theory when you're standing in front of a group of students, right? Particularly undergraduates. So there's certainly something, I think, important about being able to stand up in front of group of students and say, "Well I understand this is what you're reading in your textbook, but let me tell you how this actually works in the real world." So they say that all battle plans work until you meet the enemy for the first time, right? [chuckle] So I can tell you that

academic theory is great and I think forms an important foundation for basic decision making, but ultimately, there's something that is irreplaceable about being able to explain, in practice, how things work. Let me tell you how this stuff actually works in practice. I think it also gives you an immense amount of credibility with the students when you can say, "You know what? I've actually done this in practice, so let me tell you how you actually translate data into meaningful information when you're under the gun and you've gotta get something out to a client, and sometimes you don't want to let the better become the enemy of the good."

05:45 MF: So does the practical experience make the theory more credible?
05:48 DP: No question. No question.
05:49 MF: Okay. So there is a connection here between real world and academia?
05:54 DP: I believe there is.
05:55 MF: That's cool because most people don't think there is.
05:57 DP: Well, in retrospect, I'm not sure how you can be an effective professor without actually having done some of this stuff in practice.
06:02 MF: Right.
06:03 DP: Right? There's certainly importance to understanding the theory, but ultimately, if you can't explain how it works in the real world, I think it makes it much more difficult for the students to appreciate why you're going through all this with them.
06:18 MF: So in other words, could we say that the practical experience almost is like... Are data points that support the theory and the principles that they're studying?
06:26 DP: Right, no, I think that's spot­on. In fact, I think one of the things that's made Kennesaw State so successful in this space, in this analytic space, is the fact that every professor who is involved in the analytics program has had substantive professional experience outside of academia and so, my background obviously is in predominantly consumer finance and in risk.
06:51 MF: Right, right.
06:52 DP: But we have individuals who have substitutive experience working in hospitals, in the medical space, in manufacturing, engineering, marketing, pretty much every vertical, every domain, every discipline vertical, and they're able to bring all those experiences into the classroom.
07:10 MF: That's exciting. So along the way, in your practical experience, could you give our listeners some examples of some adverse circumstances you found yourself in? Some challenges, some obstacles to getting to where you wanted to go?
07:24 DP: Sure, so I think one of the biggest differences between what happens in theory in the classroom and what actually happens in the real world is that sometimes you just have to bring a project to fruition and it's just good enough, right? So if you've got a client deadline and you've got to get something out, you can't ultimately bring that to its final theoretical conclusion, right? Sometimes you just have to get it to 90% and you deliver what you have and you just make the best of what you've been able to put together in the time frame that you've been able to put together versus sometimes in academia people drag things out and they're willing to spend 80% of the time getting that last 20% and you can't always do that when you're under the gun with a client.
08:13 MF: That's interesting. So sometimes, no matter how much data you have, you're saying it may never be completely enough and so, at some point, you have to make a conclusion and there's some risks sometimes in those conclusions?
08:24 DP: Right, sometimes you have to just draw a line in the sand and say, "I'm just going to do the best that I can with what I've got."
08:29 MF: Okay.
08:29 DP: Right.
08:30 MF: That's interesting. So data isn't always as pure... At least the conclusions you gather from data aren't always as pure as you may want them to be.
08:39 DP: They never are.
08:39 MF: Okay.
08:40 DP: Right.
08:41 MF: Alright, well this kind of gets to something else we're very interested in, how you're building the center for Applied Statistics and Data Sciences at Kennesaw. What is your vision for the center and how are you implementing that going forward?
08:57 DP: Yeah, absolutely. So the Center for Statistics and Analytical Services at Kennesaw State serves a very important role both internally within the University as well as externally to the business community, not just in Atlanta but regionally as well. So let's start with the internal community within the University. So by way of context, Kennesaw State has 35,000 students, we have several thousand faculty, and all of those faculty have publication requirements. We have a huge portfolio of grants and contracts and all kinds of projects where we are delivering to the NIH and to the NSF and then obviously, to the local business community. All of those faculty have very strong publication requirements and research requirements, and most of those faculty also have PhD students and Master students who are working with them. The Center for Statistics and Analytical Services was really developed partially to help those faculty fulfill all of their analytical needs. So somebody may be a subject matter expert in, say, Political Science, but they're not a subject matter expert in doing statistical analysis or building prediction models, and so, that's where we come in.
10:12 MF: Okay.
10:12 DP: And we will help these professors translate this data into information to support their grants or their projects or their research. We also work very closely with the PhD programs, as well as the Master programs, supporting those students. More importantly, though, which I think really gets to your question, is how we work with the local business community. So that's really our second constituency.
10:33 MF: Okay.
10:34 DP: So the Center was chartered in part to support the needs of, like I said, the business community, both within the Atlanta area as well as regionally. We have companies who contact us almost on a daily basis, that kind of fall into one of two categories. The first is we have too much work and not enough people.
10:55 MF: Okay.
10:55 DP: And this really is a manifestation of the fact that there is a huge talent gap for people that have deep analytical skills. So we have this huge gap between the amount of data and companies all across the spectrum who are pulling in structured and unstructured data and pulling that data in is fairly straightforward, but then translating that data into something that is meaningful, that actually can support the decision­making process and support the objectives of the company, that's hard and that really is why we're seeing this huge talent gap, right? And so, our students really sit at that intersection of Mathematics and Statistics and Computer Science, and they serve partially as an extension of these teams for these companies all across the area. Again, whether you're talking about consumer finance or you're talking about risk in insurance or epidemiology and healthcare engineering, manufacturing, retail, we have almost an unlimited number of companies who are coming to us and they're saying, "We just don't have enough people who can help us translate all of this data from Twitter and from social media and from Facebook, as well as all this transaction data that's coming in, and help us improve our decision­making process." And so, our students really help fill some of those gaps for these companies.
12:22 DP: Secondarily, we have companies who come to us and it's not so much a body shop issue, it's really more of a, "I've got all this data, and I just... I don't even know what the right question is. I don't even know where to start. Can you help us take a look at what we've got and what our pain points are and what our objectives are, and help us articulate and establish what our questions should be? And then, once we've established the questions, can you then help us with the technical skills, again, to translate this massive amount of data into information to help us improve our decision­making in our business?"
13:00 MF: I would think one of your challenges might be finding the students that have both the capability of doing the data analytics, the deep technical modeling, and the kind of intuitive analytics to draw the conclusions from the data. Is that possible to have in the same person?
13:19 DP: Yeah, so it's actually an excellent question, and so, if I could kinda frame it, really, from an academic prospective, we've got students who are incredibly strong, technically, right? Who tend to come out of the sciences, right? Computer Science and Mathematics, Physics and engineering, and there's no question that these students can program and their math is unquestionable, but sometimes those students really struggle with the ability to then translate what they've done back into the context of the original business question. And I tell my students this all the time, "Nobody cares that you just did 15 weeks of mathematical callisthenics. They wanna understand why what you've done is relevant to their business problem. And so you have to be able to now put this into, dare I say, a PowerPoint presentation. You have to now put this into some type of communication that somebody who doesn't have your technical skills can ultimately understand, and they can understand why and how this affects their bottom line." And so, what we've found, certainly within the context of the Center, and more broadly within the context of the university, is that it's a lot easier to help a science student learn the importance of business acumen and the importance of translation, and the importance of interpretation and communication.
14:47 DP: It's a lot easier to help that student develop those skills than it is to take a traditional business school student, who ultimately does have that intuitive kind of wider aperture in terms of understanding the business question, and then trying to teach that student the theoretical mathematics and the programming that they need ultimately to be able to accomplish these objectives.
15:09 MF: That's exactly my point 'cause I've seen it just in our industry, in specialty finance, where we've got the talented people either on our team or on our customer's team who are really good technically, but then trying to make decisions from that data is much more difficult, and trying to communicate that data, let's say, to investors. It's much more difficult to find those same skills in the same person, and...
15:34 DP: So your point is well made, and that's something that we've been very cognizant of as we've developed all of our analytics programs at KSU. Specifically, when we launched all of our analytics programs in 2006, we did something really strange for a university. We went to the business community and we said, "This is something that we're thinking about," and back in the wilds of 2006, back in the old days, nobody had heard about this term "data science". And so I don't even know that we were necessarily using the right terms at that point, but we did sit down with many of the business leaders in and around the Atlanta area and we said, "Look, we've got this idea related to translating data into information and it doesn't fit neatly into Mathematics, and it doesn't necessarily fit neatly into Statistics, it doesn't necessarily fit neatly into the business school, but this is what we're thinking about. Can you bring us your job ads? Can you sit down with us? Tell us what your pain points are. Help us understand, if we were gonna start producing students who had a very specific set of skills, what are you looking for? How can we help close your talent gaps?" And ultimately, we are a public university, we're funded by the taxpayers of the state of Georgia.
16:51 DP: We are funded by the business community of the state of Georgia, through tax revenue. We have a fiduciary responsibility back to the state to produce talent and to produce graduates who have not just employable skills, but skills ultimately that are gonna help make Georgia more competitive. And so, when we were first launching our analytics programs back in 2006, we brought the business community up to the university, and we sat them in the conference room, and we filled out white boards, and we ultimately built our curriculum from those conversations, and it's been wildly successful. We have a 100% placement at the undergraduate and Master's level, and we haven't graduated any PhD's yet 'cause the program is still fairly recent. We're only in our third year. But we fully anticipate that, again, we're gonna have 100% placement of those guys as well. And in large part, that's because we started with the mentality of trying to ensure that we were working with the business constituency of the region and of the state to make sure that the students that were coming out were gonna be employable and could add value.
18:03 MF: So you got a PhD in Applied Statistics. Does the term 'Applied Statistics', does this mean essentially what we've been talking about, meaning using the data to apply it to a situation or an opportunity to actually use the statistics to the data? Is that...
18:20 DP: Yeah. Right. So I think that that adjective 'applied' is critical. There's a big difference between a theoretical statistician and an applied statistician. Theoretical statistics will always have a role within universities, there's no question, and we will have a need for theoretical statisticians. But ultimately, the number of theoretical statisticians we need is pretty small. What we really need is applied statisticians, and when we use that word 'applied', again, really what we're talking about is that intersection of statistics and mathematics and computer science and people who can actually extract, transport, load data from all different sources. People who can feel comfortable working with structured data and unstructured data, who can work with a variety of different programming languages, who ultimately can analyze that data and use it for the purposes of improved decision­ making. And back to your point, they also have to understand how to communicate, how to translate those results into something meaningful. So let me give you a specific example. So I teach this one class that deals with credit, a lot of credit data. So we have about 17 million cardholders, I suppose. It's all be cleansed of any personal, identifiable information, so we don't know who these people are.
19:37 DP: But ultimately, we have about 17 million cardholders, so we have about 400 or 500 pieces of information on each person. So if you can imagine, that's a matrix of about 17 million by 500. So by today's standards, I'm not sure if that qualifies as big data, but it's pretty substantive for an academic in­class teaching dataset. So in any event, these students have to go through the process, over a course of 12, 15 weeks, of really doing the equivalent of building FICO scores. So they're building risk models, trying to determine the probability of default, and then ultimately they have to find the most profitable customer segment based upon this now­found probability of default. And we have students who come back and they'll say, "Hey, I've been able to optimize the model, and it only takes me 250 predictors," and I shrug and I say, "Dude, nobody can operationalize that. That's not operationalize­able. You need to get that down to four. Give me four pieces of information that we can use." And that ultimately, mathematically, sub­optimizes the model, but ultimately it makes it monetize­able and it allows the client to actually put something into practice that they can use in business.
20:52 MF: And see, I ventured a guess, had you not had that business experience at Visa and MasterCard or AT&T Universal, I wonder if you would've been able to know that?
21:01 DP: No, it's an excellent point. I don't think that I would have. Or certainly, I would've been in a position where I didn't know what I didn't know.
21:07 MF: Yup, yup. Exactly. You didn't know what questions to ask.

21:10 DP: Right.
21:11 MF: And I love one of the quotes that I read from you in the Atlanta Business Chronicle, they've done several articles on you over the last couple of years and I was reading them over the weekend. You said, "Companies tend to be very data­rich and information­poor. Data is inexpensive and easy to capture and store, but translating it into meaningful information, that's where our Center at Kennesaw comes in." So is that the core principle or core goal of the Center?
21:41 DP: Yeah, no question. So in a... And that point is valid today as it was at the time that it came in the Atlanta Business Chronicle, and certainly, anybody that's involved in data science and advanced analytics can probably speak to that point as well. The idea here is that, again, data is cheap and easy to capture. Anybody today with some modicum of programming experience can go download literally millions of tweets, right? It doesn't take a lot of effort to go and scrap a bunch of stuff off of Twitter and scrape a bunch of stuff off of Yelp or off of any form of social media. That's easy. But then figuring out what to do with it once you've got it, that's hard, and that's where that combination of technical skills combined with some domain expertise and ultimately understanding how to write the story, all those skills come together, and that's hard.
22:43 MF: Let's talk about a very specific and controversial example, of how you applied this to a big issue in our industry, which is payday loans. And I think it was a few years ago that the KSU study, which was commissioned by the Consumer Credit Research Foundation, that you undertook had some incredibly surprising conclusions based on the data and the analytics that you did over... I guess the key question was are payday loans beneficial to subprime consumers or not? Was that the primary question? And could you summarize those conclusions? And how you got there, statistically, 'cause it shocked a lot of people because we always hear about the bad guys in payday loans, we don't know about some of the good things that happen with payday loans. And your conclusions were quite surprising. Could you share that with our listeners?
23:38 DP: Right. Yeah, I'd be happy to. So the quick background to this is I had actually done several studies for several different subprime companies, subprime lenders, not just in the Atlanta area but also regionally and nationally. And one day I'm sitting in my office, and I got a call from somebody from the Consumer Credit Research Foundation, and they said, "Hey, we've had an opportunity to look at some of your work and we have this massive dataset and we're having problems finding somebody who can work with this type of data, and we have a basic research question and we'd like for you to take a look at the data and see if you can address the research question." So their datasets, like I said, were pretty massive, but they were analogous to the datasets that I was using for the purposes of teaching, that I was just making reference to with the 17 million observations and the 400 to 500 variables. So they had multiple datasets over four years that were predominantly from a combination of Experian and payday lending transaction data. And so, in total, I think I received somewhere between 10 to 12 files and each one of these files had several million data points in it and, again, several hundred variables.
25:02 DP: So we're not talking about small datasets, and these datasets were coming in different formats and so, already you've got sort of a big data challenge in the sense that you have a massive amount of data, albeit structured, that's coming in in multiple forms. So before we can even start to get to the interesting parts about the payday loans and how do roll­overs affect people's credit scores, we have to go through several weeks of just going through the process of extracting, transporting, loading, merging, cleaning, prepping, and that took an immense amount of effort and time. Once we got the data loaded and we were able to get it into a format where it was a master file, then we really had... The core question was, "How does payday roll­overs affect somebody's financial ecosystem?"
26:01 MF: Right.
26:02 DP: That effectively was the question. And to be honest, at the time that the question was posed to me, it wasn't even necessarily posed from the standpoint of, "Hey, does this help? Hey, does this hurt?"
26:16 MF: Right.
26:16 DP: It was really positioned to me much more objectively, "What is the relationship?"
26:20 MF: Okay.
26:21 DP: Right? And so, as an Applied Statistician and a Data Scientist, I really approached it from the standpoint of Xs and Ys. I really didn't have any preconceived notions in terms of what I was gonna find. Like I said, I was really just going into this with Xs and Ys. And ultimately, what I found is that for the lowest end of the credit spectrum, so we're talking about people who have FICO scores... It was actually VantageScores, which were kind of analogous to FICO, VantageScores that were sub 600. So these are people that oftentimes can't put their hands on $17. These are the lowest end of the credit spectrum, who really are struggling to find financial resources. So in any event, at the lowest end of the credit spectrum, I found that roll­overs, payday loans, and access to payday loans was actually net positive. Not substantively positive, not in the sense that it's gonna take their VantageScores from 600 to 800, but ultimately, importantly, not negative and net positive.
27:32 MF: You mean the effect of the roll­overs on the cash flow... 27:36 DP: The effect of the roll­over on the change in VantageScore.
27:39 MF: Okay.
27:40 DP: So if you look at VantageScore, again, sort of an analogous measurement relative to something like a FICO score. If you look at somebody's VantageScore from, say, 2006 to 2007, "So how did it change? And then how did the presence of roll­overs impact that? Was that impact positive or was that impact negative?" And what I found was that it was positive, that if somebody had access to protracted roll­overs from 2006 to 2007, and then separately from 2008 to 2009, that effect was net positive. And again, I came to this with a completely blank slate and I approached this exclusively as Xs and Ys. It was just a massive data exercise from my perspective. And importantly, I made the reference to looking at 2006, 2007 as sort of one exercise, and then 2008, 2009 as a separate exercise, that was critical because of, of course, the financial meltdown that happened during that timeframe. So the objective of separating those two pieces of analysis was really to kinda control for that issue.
29:00 MF: And you use the term 'VantageScore'. Is that like a credit score for the underbanked essentially? Or...
29:04 DP: So VantageScore is a metric that was developed by the three main Credit Bureaus really as a potential alternative to the FICO score. Conceptionally it's very similar, it uses a lot of the same key pieces of information. It does... It is supposedly more representative when you start talking about people that have thin files, limited data, shorter histories. It is intended to accommodate that more accurately than the FICO score.
29:40 MF: Right. And what was the response in the marketplace and from the Consumer Credit Research Foundation to these conclusions?
29:46 DP: Sure. So when I originally did the work, it was just really presented as a mathematical exercise. So, "Here's this series of datasets, here's the basic research question, analyze the data and tell us what you found." And so, I built these general estimating equations, and did some distributional analysis, and wrote up the results and sent it to the Consumer Credit Research Foundation. And they said, "Hey, this is really interesting. We'd like for you now to put this in the context of an academic white paper," and I did. And we loaded it up on the Social Science Research Network, which allowed me to then get feedback from the academic community in advance of ultimately taking it to a peer­review journal. And as I'm sure you can imagine, there was a lot of feedback, really, from both sides, and some people thought, "This is really interesting. I'd like to better understand the math behind that."
30:49 MF: Right.
30:49 DP: And of course, I was happy to have these conversations. Several people asked me for my programming code, and I was happy to give them that as well. As an academic, I'm all about replicating research and replicating results, so I was happy to make my math available, happy to make my code available. There were some other people that took a look at it and, rather than engage in a mathematical­scientific conversation, were much more interested in maybe dismissing the results because it didn't necessarily fit within their expectations.
31:23 MF: Right.
31:23 DP: And so, that was frustrating, but, for the most part, the conversations were engaging.
31:29 MF: Well, terrific. And also, in the Business Chronicle, there was an article last year about how you're expanding partnerships for the Center.
31:39 DP: Yup.
31:40 MF: Including such companies as AT&T, Southern, Home Depot, Equifax, which is sort of new, I think, for academic institutions. Could you comment a little bit on how you've been developing these partnerships? And does this kind of tie into your belief that you've got to apply the data, you've got to use it, make data information, not just analytics? Is that the connection?
32:03 DP: Yeah, absolutely. So if we can kind of open the aperture a little bit and talk more broadly just about university/corporate partnerships. Going back to the point that I made earlier that ultimately we are a public university and we're funded by the taxpayers, right? And so, we have this responsibility to make sure that the people that graduate from Kennesaw State are able to go back into the business community within the state of Georgia and add value, right? And so, to that point, we need to constantly be in contact with the ultimate constituents or, if you will, our customers of our products, kind of with quotes around that, right?
32:48 MF: Right.
32:48 DP: Ultimately we are producing a product. Not to be crass, but ultimately that product is graduates of the university and we need to make sure that the moneys that the tax payers of the state of Georgia are investing in Kennesaw State is creating a return on that investment by graduating people who have meaningful degrees, who can add value and can walk into the business community and, again, help contribute to the economy of the state. So to ensure that we're able to do that regularly, we check in with the business community, and certainly within Atlanta, you've got kind of the usual suspects, many of which you named. We also have a lot of small companies, a lot of small startups that we do a lot of work with, who are regularly at the university, and we're constantly checking in with them and making sure that what we're doing is ultimately meeting their needs. And so, back to how this works with the Center, so if we're going to continue to graduate people that have skills that are in demand in the marketplace, the raw resource that we need from an analytics perspective in order to continue to fulfill that objective is, well, data, and it needs to be real data because I can tell you that academic data is effectively worthless.
34:06 DP: Academic data that comes out of textbooks is 100 observations and three variables, and it all fits nicely into an Excel spreadsheet, and everything's clean, and everything generates these incredibly strong relationships, and the students pretty much just have to point and click on a couple buttons. And if that's the only data that a student sees prior to graduation, then we're committing academic fraud. Right?
34:28 MF: Yeah. Sure. It's not useful.
34:29 DP: Because that's not real world. And so, if we're going to ultimately create and produce graduates who are able to walk into the Fortune 500s that are based in Atlanta, then it needs to be a symbiotic relationship. So those companies are gonna have to give us some data that we can use in the classroom, obviously purged of any personal identifiable information and stripped of any competitive information or anything that they deem to be competitively important, but if we can get real data, real messy hairy complex difficult data, and use that for the purposes of teaching in the classroom, that creates real­world skills, right? That creates meaningful skills that then we can graduate and put them back out into the marketplace here locally, and ensure that we're continuing to make Atlanta and the region economically competitive. And to that point, we hear this... This trend now is all about buying local and eating local, and I would just add to that and I would say hire local. There's no reason why... Anybody that's looking for data science talent in Atlanta, there's no reason why they should be trying to import exotic kids from California.
35:52 MF: Right.
35:52 DP: You don't have to go to the West coast to find data science talent because ultimately, if a kid has grown up in California and their family is in California and more importantly his girlfriend is in California, I don't know how long, ultimately, they're going to wanna be here in Atlanta, right? But if you can find kids from Kennesaw state and from Georgia Tech and Georgia state who are from Atlanta, they've gone to high school in Atlanta, their family is in Atlanta and they have the skills, they're gonna stay in Atlanta. So eat local, buy local, hire local.
36:28 MF: And again, it sounds like the most successful ones are those who can combine the theory and the application of the data, even though they may not have practical experience, but it's information practically applied.
36:41 DP: Well, so let's talk about that practical experience for a minute, Michael. So you were talking about how the Center fits into all of this. The Center actually is a very important source of practical experience for the students. So when they make me the benevolent dictator of education, [chuckle] I'm going to require that every undergraduate student is required to have some type of vocational training as part of their discipline. Go out, get the internships, go out, have the co­ops, do the ad hoc project work. Even if it's unpaid, just go and get the experience, right? And so, for the small percentage of students at KSU who can't get internships, which there aren't many, we actually do an excellent job of ensuring that the students have internship experience, but to subsidize that, we see a lot of students who will come through the Center and they do project­based work through the Center. And so, the majority, if not pretty close to all of the students at the undergraduate, master's and PhD level who are coming out of the analytics programs at KSU have some kind of practical applied experience with real world data, either through the context of an internship or through a project or through the Center.
37:55 MF: Wow! Now, we only have a few minutes before we wrap this up, but I think our listeners would like to know a little bit more about the woman behind the Center, the woman behind the PhD. And I know from our friendship at the Cathedral of Saint Philips that you've... And I've been a student in the adult Sunday School classes that... You speak on a lot of interesting topics, and the one that I think was most relevant today that I'd like you to comment on was faith and big data, and you talk about volume, variety, and velocity. And you even bring your grandmother into the story and I think it would help our listeners understand that there is also a spiritual dimension behind your work. So could you quickly summarize your thoughts on faith and big data?
38:38 DP: Yeah, it's interesting. So I talk a lot. [chuckle] Well, I talk a lot, but I actually go out and I speak to a lot of different audiences on the topics that we've discussed here related to big data analytics and, "Is the data that we work with today really substantively different," and, "What does the educational landscape look like?" And it's interesting, when I go to a lot of these companies and I do a lot of these talks, it's almost every single time, somebody will walk up to me either before or after and say, "I heard your talk, your podcast on faith and big data, and I really enjoyed it." And I think, "Wow! Of all the sort of intellectual machinations that I'm involved in and all of the academic papers and talks that I've given, that's actually the one people wanna talk about, is faith and big data." So, just by quick way of context, when we hear this term "big data"...
39:35 MF: What does that mean? Our viewers... Our listeners may not quite understand what big data means.
39:39 DP: Yeah, yeah, so maybe you could have me come back and I could do a whole piece on big data.
39:45 MF: Okay. Alright. We'll do that.
39:46 DP: But the short version of this is that big data is really defined as sort of this intersection of volume of data, the size of the files, but it's not just volume because if it was just volume, then we would've always had a big data challenge, in the sense that, back a million years ago, I had a PC Junior when I was in high school and I think that PC Junior had like 64K of memory. [chuckle] We don't even have anything anymore that's 64K and so ultimately, if somebody gave me a file that was a 128K, I would've had a big data problem, right? So it's not just about... It's not just about volume, it's not just about the size of the files, it's the intersection of the size of the files combined with the velocity of the data. So we talk about things like Internet of Things, IoT, or other sources of data. I made reference to Twitter and social media. The fact is that that type of data is being generated all the time. All the time. And so, you get this tsunami of data that's coming in every second of every minute of every hour of every day all the time. Combined with the fact that it's big, that it's big volume. But then also you've got this overlay, this interaction issue of variety. So data doesn't come to us typically anymore as rows and columns. Data is coming to us as audio files and it's coming to us as pictures, pixels. And it's coming to us in text files and these crazy types of unstructured files. And like I said, being able to scrape social media stuff.
41:33 DP: So you've got this combination of volume and velocity and variety altogether that is creating the problems, or the challenges, that we now see that we've termed big data. And so, in the context of this podcast, what I made reference to is the fact that the greatest number of "religious designations" that... Or faith­based designations that we see now in this country is none, right? That people just aren't identifying with a faith. And so, in the context of this podcast, I really compared and contrasted this to my grandmother's time. And so, my grandmother was born in 1901 and she was a second­generation Irish immigrant and she was in Elwood, Kansas. And I talk about the fact that if I look at my life, defined today by volume, velocity, and variety, and I compare and contrast that to my grandmother's time and she had her own challenges with volume, velocity, and variety, right? But I use that as a basis to kind of explain why we went from 70%, 80%, 90% of people in my grandmother's time who attended church on a regular basis and really defined themselves in the context of a faith, to where we are today, where the majority of people do not define themselves as having a faith. And so, just briefly, in the context again of velocity, and volume, and variety, the volume's probably pretty obvious to people, right?
43:06 DP: We talk about the volume of stuff we have to do and a lot of people kind of point to that as a basis for why people don't go to church anymore. And my argument is, just like with big data, it doesn't have anything to do with volume because my grandmother's time, she didn't have the same type of volume that I do or the same types of things that were contributing to the volume, but she was a teacher and she had to get to school an hour early because she ultimately had to start the fire in the fireplace to make sure that the schoolroom was heated before the kids got there. And for a lot of those kids, she had to make the lunches and then she had to teach the classes. And then when she got home, she actually had to make sure that she had knitted the blankets to make sure that her own children were warm for the wintertime. And so, she had her own challenges related to volume, they were just different from my own, and so, I don't know that that's the right thing to point to in terms of what we now call is the none. Variety is another point that I make reference to in this podcast. So if you look at kind of variety of faith, right? So in my grandmother's community, everybody was an Irish­Catholic immigrant. Everybody went to the local Catholic church.
44:16 DP: There was just no... There was no variety in her space and in her community. And there are pros and cons to that, right? So in my own community, my children are blessed to be going to school with kids who are Hindu and Muslim and Jewish and Christian and across a wide variety of faiths, but then the challenge there is that they don't go to church with the same kids they go to school with, right? There aren't too many kids who are Episcopalian in their class, but the ultimate reality is that they're blessed and I think that they are stronger in their own faith because they're able to walk other kids' faiths, right? And so, there's a benefit there to the variety, but I don't know that that's necessarily something that is indicating the nones either. And then the last one is just speed, right? The velocity. And I do think that that's something that's critical, that is fundamentally different in our space relative to my grandmother's space. And in that podcast, I really make reference to that fact that I can remember, as a child, going to my grandmother's house on a Sunday and she really created and observed Sabbath, right? So we would go to their house and we would go to church and then we would come home and the television would not be on, right? There's no television, there's no radio.
45:46 DP: And I remember having to sit at the table and having to snap beans and having to listen to my mother and my grandmother talking, and as a eight, nine­year­old kid, it was so boring, right? But there is... I reflect on it now and it was actually incredibly beautiful and created this very special, kind of spiritual foundation. And I don't know that my grandmother was doing that on purpose, I think it was just inherent and intrinsic as to how she lived her life, right? That you slow down and you have to create that space for Sabbath. And I think that that's something that we struggle with, particularly, in our household, and I would argue within the context of our community, that slowing down and really embracing a Sabbath, and that Sabbath doesn't necessarily have to be present on Sundays as we're running to lacrosse practice and soccer practice and everything else. It could be on Tuesday evenings or it could be on Saturday mornings or whenever it is, but just that idea of slowing down is something that you really have to make time for.
46:51 MF: That's a wonderful, wonderful example, Jennifer, of how you, again, are connecting data and applying it to other things, whether it's business decisions or your faith. That's really cool with the volume, variety, and velocity that defines big data, you're applying it to your spiritual life as well and to me, that's... You're connecting the dot... I'm connecting the dots of this person we know as Jennifer Priestly, both a PhD and a Sunday School teacher and a mother and the director of the Center for Statistics and Analytical Services at Kennesaw State. It is so exciting how you've connected all this in your life and we're gonna have to wrap it up this morning, but it's been a wonderful, wonderful 45 minutes. Do you have any final thoughts on kind of what you've learned in this journey and building this Center at Kennesaw?
47:48 DP: It's been wonderful and I can't speak highly enough of the students at the University. I go to cocktail parties and things and people oftentimes will make reference to the millennial generation and how they don't work hard, and how they have the sense of entitlement, and that... I don't know, that may or may not be true based upon their specific experiences, but I have to say, the kids at KSU, they work so hard and they have such an incredible work ethic. And we hear that over and over and over again from companies, and that's why we have 100% placement rate. The kids are smart, but ultimately, they have a work ethic, and they come to work and they put their all into it every single day and we see that in the classroom. And it's like when you send your own kids off to somebody else's house, you don't ultimately know how they're gonna behave, but...
48:38 MF: Right. Same thing here. Yeah.
48:39 DP: Yeah, exactly. And so, much like a proud parent, when companies come back to us and say, "You know what? I'd like to hire another kid like this one." It's great. Ultimately, that's what makes me go to work every day.
48:53 MF: So is that the most important metric of success for the Center that you're director of, essentially? Or one of them?
49:00 DP: Just the whole idea of interacting with the local business community to bring the data in, make sure that what we're teaching is aligned with the market place, make sure that these students are developing the skills that they need to contribute to the economy, to contribute to the business community, and then ultimately, have those students come back and hire more students, as Alumni coming back and hiring more students. That's what it's all about.
49:28 MF: I can't think of a better way to conclude our discussion this morning. Thank you very much, Dr. Jennifer Priestly.
49:32 DP: Thank you, Michael.
49:35 IS: Thank you for joining Michael Flock and his guests on the Capital Club Radio Show. For more information on future interviews, please visit us at This program is brought to you by FLOCK Specialty Finance, where clients are provided knowledge and insights to help them grow their business in complex and risky market. FLOCK is more than a transaction.

Page of

Download 63.18 Kb.

Share with your friends:

The database is protected by copyright © 2023
send message

    Main page