Episode 13: Breeding Data Science Unicorns
Download MP300:00:01 Dr Genevieve Hayes
Hello and welcome to value driven data science brought to you by Genevieve Hayes Consulting.
00:00:06 Dr Genevieve Hayes
I'm doctor Genevieve Hayes, and today I'm joined by Doctor Peter Prevas, a man who aspires to become known as the Willy Wonka of the water industry.
00:00:17 Dr Genevieve Hayes
Peter is a civil engineer and a social scientist who manages the data science function at coliban water in regional Australia.
00:00:25 Dr Genevieve Hayes
And runs leading courses in data science for water profession.
00:00:29 Dr Genevieve Hayes
He is also the author of a number of books, including Principles of Strategic Data Science and the recently released data Science for water utilities.
00:00:40 Dr Genevieve Hayes
Peter, welcome to the show.
00:00:42 Dr Peter Prevos
Genevieve, it's great to be here.
00:00:44 Dr Genevieve Hayes
It's fantastic to have you.
00:00:46 Dr Genevieve Hayes
Data science Unicorn is one of those terms that's been bandied about quite a bit over the last few.
00:00:52 Dr Genevieve Hayes
Years in your books you describe data science unicorns as those people who sit in the middle of the data science Venn diagram.
00:01:02 Dr Genevieve Hayes
And I'm sure most of our listeners have seen this before. This is the one that describes the data science skill set as being the intersection of domain knowledge, mathematical.
00:01:13 Dr Genevieve Hayes
Skills and computer science skills. Does that sound?
00:01:17
Right.
00:01:18 Dr Peter Prevos
Yeah, that's that's the model I'd really like to use quite a lot to to explain the other science, but it's also a little bit controversial as you as you are alluding to.
00:01:27 Dr Genevieve Hayes
Oh yeah, I I can imagine. But it's it's a nice like all models. I think you said the other day you're quoting George Box.
00:01:34 Dr Peter Prevos
Yours box, yeah.
00:01:35 Dr Genevieve Hayes
All models are wrong, but some models are useful.
00:01:38 Dr Peter Prevos
Exactly. And that's that's something to live by as a, as a engineer dealing with practical issues every day.
00:01:44 Dr Genevieve Hayes
But as you'd expect, the people who sit at the centre of that Venn diagram, they're pretty rare. In fact, some people would even go so far as to describe them as mythical, hence calling them.
00:01:55 Dr Genevieve Hayes
Econs and in this episode we're going to look at approaches organisations can take to developing such individuals from within or reading data science unicorns as Peter would would call it. But looking at your background, Peter, I'd say that you yourself could be described as a Unicorn.
00:02:15 Dr Genevieve Hayes
In addition to your years of experience in the water industry, you also hold a bachelor's degree in engineering and MBA and a pH.
00:02:24 Dr Genevieve Hayes
D in marketing, I suspect there aren't too many people out there with that combination of skills and qualifications.
00:02:32 Dr Peter Prevos
And that was exactly my my point of of saying let's breed quote, unquote, data science unicorns. A lot of the written material about these mythical creatures of data science unicorns comes from people who.
00:02:45 Dr Peter Prevos
Start from the computer science and statistics point of view. From a generic perspective and and also it seems like some people think that data science is this new.
00:02:54 Dr Peter Prevos
But it's not. I've been analysing data for my whole 30 year career, but early in my career I had a pencil, a calculator and a piece of paper, and my data was thousands of pieces of paper and I just leave.
00:03:05 Dr Peter Prevos
A file and and so I've seen that whole evolution of of data. And when I looked at that Conway Venn diagram I.
00:03:13 Dr Peter Prevos
I sort of recognise myself in there because yes, I have statistical and mathematical knowledge through my engineering degree. I am a subject matter expert because I'm a water engineer.
00:03:22 Dr Peter Prevos
That's that's who I am.
00:03:23 Dr Peter Prevos
And always had a bit of an affinity with.
00:03:25 Dr Peter Prevos
Computer science because I started.
00:03:27 Dr Peter Prevos
Writing computer games in the 19, my little.
00:03:29 Dr Peter Prevos
Atari 8 bit.
00:03:29 Dr Peter Prevos
Computer got into got into machine language.
00:03:31 Dr Peter Prevos
And so I had a bit of an affinity with computer science, and when I wrote my PhD, I found out very quickly that Excel couldn't meet my needs because I've I've developed thousands of spreadsheets, whole jungles of interconnected.
00:03:43 Dr Peter Prevos
Spreadsheets to to do what I need to do, but as I got into more advanced statistics to do my research, I decided to use the R language and then my computer science skills started kicking in or my my amateurish computer science skills. And yeah, started to combine those. So that's why.
00:03:59 Dr Peter Prevos
I believe to to breed data science uniforms is to start from people with a certain subject matter expertise to teach them computer science and together with the real quote unquote computer scientists, I think that that's forms that forms a great team, but the subject matter experts need to understand the computer science so we get away from the fear of the black box.
00:04:21 Dr Genevieve Hayes
And also just how understanding computer science creates a lingua franca between the two.
00:04:27 Dr Peter Prevos
Ideally it would, but people have this fear of of computer science.
00:04:31 Dr Peter Prevos
I'll bring my own computer to work because I like working in Linux and Emacs, and my my colleagues look at me really.
00:04:39 Dr Peter Prevos
Ohh, why do you use amados? You know so that that that they have this idea that it that it's that's doing stuff in the computing way. Because I write I write presentations in more than.
00:04:51 Dr Peter Prevos
And my colleagues think that is backward, but I'm actually saying no, it's it's forward thinking it's let's embrace computer science and and the text editor and start writing code. And that's what my course. And then subsequently the book is all about to.
00:05:04 Dr Peter Prevos
Teach people who are subject matter experts to lose that fear of fighting code.
00:05:09 Dr Genevieve Hayes
We were using S plus for statistics when I was an undergrad, and as you know, S Plus was the forerunner for R.
00:05:17 Dr Genevieve Hayes
And back in those days, there didn't exist a Windows version of S Plus. So in order to use it, you had to have a Linux computer, so it involved going into this Linux lab on campus.
00:05:31 Dr Genevieve Hayes
And I think there were like 2 Linux labs on the whole of the ANU campus, and I remember thinking.
00:05:37 Dr Genevieve Hayes
Why am I using this horrible antiquated system?
00:05:41 Dr Genevieve Hayes
And now I actually have my own Linux computer and I prefer using it to the Windows computer.
00:05:46 Dr Peter Prevos
And it's it's it's a realisation about the awesome power.
00:05:49 Dr Peter Prevos
It gives you to.
00:05:50 Dr Peter Prevos
Just write in plain text and I'm I'm running in the next series of courses now and the the part that I always enjoy is the first session that I've run and I have some slides which is a an image of a real spreadsheet that I had to reverse engineer to make some changes and the spreadsheet had been around for probably a few decades and I had to start drawing arrows on and and trying to figure out how all these.
00:06:11 Dr Peter Prevos
The cells will link to each other and then in the next slide I actually show the students who don't know any code yet what it looks.
00:06:19 Dr Peter Prevos
Blocking code and and I'll and I'll walk him through it as a as a logic to say here's a story that's written for you that I can follow rather than arrows pointing everywhere.
00:06:28 Dr Peter Prevos
And I do it in RI do it in Python And I do it in Julia to show that they're all the same.
00:06:33 Dr Peter Prevos
Really, it's just they're semicolon goals in a different spot. Yeah. And and that's I think that.
00:06:37 Dr Peter Prevos
Opens a few eyes that to to.
00:06:39 Dr Peter Prevos
Remove the fear for code and to show that you're really writing a story about the steps of your analysis.
00:06:46 Dr Genevieve Hayes
What was that term you used in your book? Literate coding, is that right?
00:06:51 Dr Peter Prevos
Literate programming? Yeah, that's that's that's the the next step, which is what I used to write the data science book is where you combine text and code together and the various systems and sweave when you write in latex or there's our markdown, which is very popular.
00:07:07 Dr Peter Prevos
I use org mode in Emacs, but the beauty of that is you can write your prose in the markdown language and then you have computer code and then you can choose that you don't include the source code or not, or include the output or not. And so it's it's ideal to write code that documents itself.
00:07:26 Dr Peter Prevos
And it's very popular among our programmers.
00:07:28 Dr Peter Prevos
It's it's a bit. It's a bit like Python. What's it called in Python? The Jupiter? That's it, yeah.
00:07:32 Dr Genevieve Hayes
Jupyter notebooks.
00:07:34 Dr Genevieve Hayes
Yeah, it's very popular among data scientists in Python as well.
00:07:38 Dr Peter Prevos
Yeah, and it's literal programming is a term by by the goal of computer science, Donald Knuth.
00:07:43 Dr Peter Prevos
Oh yeah. So he came up.
00:07:44 Dr Peter Prevos
With that way, before the technology existed.
00:07:46 Dr Peter Prevos
Or even implemented.
00:07:47 Dr Genevieve Hayes
Yeah, I'm. I'm fascinated by those computer scientists who come up with ideas before it's even possible. It's like Alan Turing.
00:07:54 Dr Genevieve Hayes
The idea of a touring machine, just it just blows my mind because this guy came up with it before what we consider to be a computer even existed.
00:08:03 Dr Peter Prevos
Yeah. And doing that just in his mind or with pieces of paper, just just and that's why that's what I love about the history.
00:08:08 Dr Peter Prevos
Of I watch a lot of YouTube.
00:08:10 Dr Peter Prevos
Just about across the old engineers who were involved in the 70s and the 60s, and the list the list language and those sort of things because it really teaches you the principles of what a computer is really.
00:08:21 Dr Peter Prevos
And a lot of people have lost that because they're just clicking on the screen with a mouse and talk about magic Windows and Apple.
00:08:28 Dr Peter Prevos
They like magic tricks because they show you a world that doesn't exist. It's illusion that you have a file that you drag from one place to another, but what it does, it actually obfuscates what the computer really.
00:08:37 Dr Genevieve Hayes
Does. Oh, that's the reason why I prefer Linux to windows, because.
00:08:42 Dr Genevieve Hayes
It removes some level of that obfuscation.
00:08:46 Dr Peter Prevos
Exactly right. Yeah. You're actually computing. You're not clicking on buttons. I enjoy computing, and I'm. I'm trying to get that passion back.
00:08:55 Dr Genevieve Hayes
Yeah, a friend of mine or former colleague of mine once said the difference between Windows and Linux is with Windows.
00:09:02 Dr Genevieve Hayes
He he was asking himself, can I do this task in Windows whereas with Linux he was asking himself, OK how can I do this task in Linux?
00:09:13 Dr Peter Prevos
That's a great example, yes.
00:09:15 Dr Peter Prevos
And people are scared of Linux. I'm trying to get my business here, instal the Linux server you know on cloud or on premise and they just scared of it.
00:09:26 Dr Peter Prevos
So I've had. I've had several meetings, but I I want to move to the professional studio version and it only runs of security and we have no.
00:09:36 Dr Genevieve Hayes
I was terrified of Linux until I worked at a workplace where the majority of people use Linux, and then I really tried to learn it and now it's like this is so much better than Windows.
00:09:46 Dr Peter Prevos
I know.
00:09:48 Dr Peter Prevos
Yeah, yeah, I've been using Linux for 25.
00:09:50 Dr Peter Prevos
Years and I would never go back to it.
00:09:52 Dr Genevieve Hayes
To go back to what we're talking about before.
00:09:55 Dr Genevieve Hayes
Given your background in engineering, what led you to also pursue qualifications in a social science discipline, such as marketing?
00:10:04 Dr Peter Prevos
So there's an interesting anecdote for that, so.
00:10:07 Dr Peter Prevos
Just after I finish my engineering degree, I did an ask degree just for fun.
00:10:11 Dr Peter Prevos
And when people ask me why I did it and I and I majored in philosophy, it took me 10 years to do because my answer was I like studying useless things, and I just. I just lost my development.
00:10:21 Dr Peter Prevos
But then I realised, well, I want to progress my career. I should do this thing called an MBA one stage and then my employer had a really good offer on on helping funding.
00:10:31 Dr Peter Prevos
It so I just jumped into it and it was a regional MBA at Latrobe University, so it.
00:10:36 Dr Peter Prevos
Was here in.
00:10:36 Dr Peter Prevos
Bendigo weekend, actually, so that was a perfect opportunity and I just jumped into it.
00:10:40 Dr Peter Prevos
My very first subject was marketing by the late Professor Brett Walker and he asked everybody, why do you do marketing?
00:10:48 Dr Peter Prevos
My answer was it's a compulsory subject. I'm an engineer on a new marketing because that's the sort of answers I gave, but red was very gracious and he said OK, I'll convince you that you're and he was the most inspirational lecture lecture I've ever had. And going through the MBA and sort of what did a few research subjects I enjoyed doing research.
00:11:08 Dr Peter Prevos
And although I always thought that somebody with a PhD is somebody who is very good at something really unimportant, like very.
00:11:13 Dr Peter Prevos
Because you have.
00:11:14 Dr Peter Prevos
To be very focused, I thought maybe I.
00:11:16 Dr Peter Prevos
Should do lecturing.
00:11:17 Dr Peter Prevos
And then the professor said, well, you have to do you have to have a PhD or at least do one. So how how how, how can that be? So I'll just jump into that.
00:11:26 Dr Peter Prevos
And and I picked the subject of buckling, but I combined it with management. So my, my, my subject area customer centricity cause within my industry at that time people started talking about being customer centric rather than behaving like a government entity.
00:11:41 Dr Peter Prevos
So it was a good research topic.
00:11:42 Dr Peter Prevos
There was, you know, nothing had been.
00:11:44 Dr Peter Prevos
Done before and. Yeah, and.
00:11:46 Dr Peter Prevos
And through that I got in touch with the data science because one of my colleagues who was not one of my staff members, Jenny said to.
00:11:54 Dr Peter Prevos
Me. Hey, you should look at this thing called data science. I think you'll find it. Really.
00:11:58 Dr Peter Prevos
So this was probably about 2010 or 11 and thought, oh, data in science I.
00:12:02 Dr Peter Prevos
Never thought you could put those.
00:12:03 Dr Peter Prevos
Yeah, and did a did an online course through Coursera with John Hopkins University and fell in. Love it.
00:12:11 Dr Genevieve Hayes
Obviously, had you not done that marketing degree, you probably would never have had that conversation that led you to you to discover of data science at that time.
00:12:20 Dr Genevieve Hayes
But do you think that what you learned as part of your marketing degree has made you better data scientist?
00:12:27 Dr Peter Prevos
Absolutely. Because because I'm an engineer, which is fully the physical sciences. So I've I've fully embraced that. I understand physics and I can do all that sort of mathematics through my arts degree.
00:12:38 Dr Peter Prevos
I already started getting affinity but also qualitative side of life cause not everything can be fully expressed in numbers, but I didn't really have a grasp on what that means.
00:12:47 Dr Peter Prevos
Now in my marketing is this nice grey area between quantitative and qualitative, and there's a lot of really good.
00:12:55 Dr Peter Prevos
Statistics out there to to how do you look at a customer survey, for example. So and that's that's sort of fascinating me and I and I researched in.
00:13:03 Dr Peter Prevos
My dissertation and.
00:13:04 Dr Peter Prevos
The thing that I'm most proud of of my dissertation is really the the part where I'm merged qualitative and quantitative aspects of data and using data science to network analysis.
00:13:15 Dr Peter Prevos
For example, I did try some language modelling but didn't quite work hard, but yeah, so that's how I started using those things, but not necessarily quantifying the qualitative side but still recognising it, but expressing in the statistics of factor analysis to cluster analysis those sort of technique.
00:13:31 Dr Peter Prevos
I want to Start learning these techniques. That's that really got me really got me interested was that went beyond the typical engineering, which is all time series and.
00:13:39 Dr Genevieve Hayes
How did you incorporate network analysis into your analysis?
00:13:43 Dr Peter Prevos
Yeah, that that was an interesting little inspiration.
00:13:46 Dr Peter Prevos
I had to.
00:13:47 Dr Peter Prevos
My problem was I'm a I'm a water expert and that was a problem in my research because.
00:13:52 Dr Peter Prevos
Just you can't write a dissertation and and from your own experience obviously, so I had.
00:13:57 Dr Peter Prevos
To quantify that, So what I did, I started doing some advanced Googling to find all the academic papers about the water industry that mentioned customers in their different forms. And then I downloaded all the abstracts.
00:14:12 Dr Peter Prevos
First I tried some text analysis on that some the topic analysis, but it didn't really give me anything that I could interpret.
00:14:19 Dr Peter Prevos
So then what I decided to do is manually code these abstracts in topics.
00:14:24 Dr Peter Prevos
And I worked out a methods to then create a network of these topics. So if a paper mentions 2 topics, that's a little bimodal network and another paper mentions the same topics plus another one and then another one.
00:14:36 Dr Peter Prevos
And if you put it all together, you get quite an interesting diagram of the state of knowledge within that in this, in this case about customer centricity.
00:14:44 Dr Genevieve Hayes
So you're building a knowledge graph of all the information.
00:14:47 Dr Peter Prevos
That's the word knowledge graph, yes.
00:14:49 Dr Genevieve Hayes
I actually had a guest on Alessandro ***** who talked about knowledge.
00:14:54 Dr Peter Prevos
Yeah, I love to look. I learn more about that cause I sort of intuitively invented this method and it worked out really well because it allowed me to say this is what the industry knows or doesn't know about this topic, rather than relying on my own expert.
00:15:08 Dr Peter Prevos
I also use it for I did interviews as well and then for the interviews use topic analysis. So let's manual.
00:15:15 Dr Peter Prevos
Labelling, not not automated. Also, just visualise that in the network because it was just the easiest way.
00:15:22 Dr Genevieve Hayes
In the preface of your new book, data science for water utilities, you state that your motivation for writing this book is to breed data science unicorns by introducing water professionals to using code to solve problems. What inspired that motivation?
00:15:39 Dr Peter Prevos
What inspired the motivation was and my experience through my dissertation and starting to analyse data in your language instead of instead of spreadsheets through that experience, fully converted code is superior to to a spreadsheet for anything that's repetitive.
00:15:53 Dr Peter Prevos
Also, what is happening in the industry? There's a lot of interest in using data science food in my industry.
00:16:00 Dr Peter Prevos
But the problem is there are probably more solutions than there are problems. So as all these vendors out there trying to sell all this stuff and they talk about AI and machine learning and they just mix up the terms without any consideration, what it really means and and they're just presenting it as a as a magic trick. Again, let's use it as a theme for this.
00:16:21 Dr Peter Prevos
And the problem is a lot of engineers, specially the older ones are like I don't want a black box because we.
00:16:27 Dr Peter Prevos
Want to know how the thing works, right?
00:16:28 Dr Peter Prevos
You know, the thing about engineers, engineers are the people who take their toys apart and then to to see how they work.
00:16:35 Dr Peter Prevos
And the problem is that we can only really have a conversation with vendors if we understand how it works, we can ask critical questions to these vendors. So that's one motivation. So Even so I'll tell my students even if you.
00:16:48 Dr Peter Prevos
Don't end up writing code. I hope that from this you can at least be able to have a conversation with the data scientists and and take some of their mythology out of them.
00:16:58 Dr Peter Prevos
Unfortunately, a lot of these vendors you're asking about how the product works. They they seem to hide behind IP, but they don't tell you that what they do is all based on open source algorithms.
00:17:09 Dr Peter Prevos
I think as an industry for us to embrace these new ways of analysing data, we need to understand this.
00:17:15 Dr Peter Prevos
So I'm I'm able not to ask questions when when the data side firm comes along and says hey, we can do this then the other so well, how does it work? What's what's what's the algorithm that you used? What sort of training models have you use?
00:17:28 Dr Peter Prevos
And and also engineers within word utilities and other professionals analysed a lot of data in spreadsheets and my my task through my team is to help my colleagues doing that we use file BI but it's still a lot of spreadsheets being used and I've already convinced all my colleagues to start writing our code that they do.
00:17:48 Dr Peter Prevos
Water quality analysis for ad hoc things.
00:17:51 Dr Peter Prevos
It's just a it's just a better way of doing things and then we can share scripts and and we can critique because they're a lot more, a lot more resilient than it's.
00:17:59 Dr Genevieve Hayes
What you said a few moments ago about all these tools that you can use for water analysis, how the salespeople hide behind IP, but they're really open source under the hood.
00:18:10 Dr Genevieve Hayes
I've found if you know the particular packages for performing a particular type of analysis well enough, you can actually figure it out just by the output.
00:18:19 Dr Peter Prevos
Yeah, I've. I've done that to a degree and there's there's an interesting development. So one of the things in in sewer networks, so one of the problems for sewer networks is roots growing into it and they start locking up and they collapse because they're old.
00:18:31 Dr Peter Prevos
And one of the things that water utilities do, we send the little.
00:18:34 Dr Peter Prevos
Robot camera in there and assess the.
00:18:36 Dr Peter Prevos
Condition. That's the unenviable task of every graduate engineers to go through hours and hours and hours of footage, and then we use it to find cracks, et cetera. Now there are now obviously image recognition is a perfect tool for.
00:18:49 Dr Peter Prevos
And there are companies that sell that software. But really what they're selling is the training data. We helped them train.
00:18:56 Dr Peter Prevos
Data because the algorithm is almost off.
00:18:59 Dr Peter Prevos
The shelf, there's.
00:19:00 Dr Peter Prevos
A colleague at another utility in Melbourne said in a weekend they slap something together that could almost do it.
00:19:07 Dr Peter Prevos
So really what these companies are selling is is the.
00:19:09 Dr Peter Prevos
Training the data.
00:19:10 Dr Peter Prevos
Is the and then the algorithms are.
00:19:12 Dr Peter Prevos
Open, but let's.
00:19:13 Dr Peter Prevos
Just have. Let's just be open about these things.
00:19:15 Dr Peter Prevos
And have real.
00:19:16 Dr Peter Prevos
Real meaningful chats with these.
00:19:18 Dr Peter Prevos
People about how they can improve as.
00:19:20 Dr Peter Prevos
Well and solve real.
00:19:22 Dr Genevieve Hayes
I've actually seen tours that, you know, do things like generic image recognition, translation, transcription, all that, and they are just using either the AW, S, Microsoft or Google API's under the hood, and I can actually spot which ones which, because I know three sets of APIs well enough.
00:19:41 Dr Peter Prevos
That's and that's and that's fine. But just let's see it so. So my idea for the data science data sounds Unicorn.
00:19:47 Dr Peter Prevos
It's not that they're lying, but, but we need to have the. So that's the the two the the two reasons are yeah, be more informed than be informed in the market, but also.
00:19:58 Dr Genevieve Hayes
On YouTube, there's a video of a workshop you delivered last year entitled Data Science Unicorn breeding programme teaching coding skills toward professionals.
00:20:07 Dr Peter Prevos
What's that on YouTube?
00:20:08 Dr Genevieve Hayes
Yep, it's on YouTube, so clearly this isn't something that you just aspire to do. It's clearly something you've put into practise already.
00:20:17 Dr Peter Prevos
Yes, it started in 2019 or 18 even. So there's an organisation got water research Australia and we remember and and I mean we were always active and doing events and stuff and we started.
00:20:29 Dr Peter Prevos
With the idea.
00:20:30 Dr Peter Prevos
Was I felt a bit lonely and I said I love to talk to.
00:20:33 Dr Peter Prevos
Other people who.
00:20:33 Dr Peter Prevos
Were who have the same interests as me within my industry and develop these skills?
00:20:38 Dr Peter Prevos
So developing a community of interest.
00:20:41 Dr Peter Prevos
But I couldn't find enough people to to that were at the same level as I was, so I decided I want to do a course, create my own friends, if you like.
00:20:50 Dr Peter Prevos
And that course grew and grew. And yeah, I've done workshops now and and and talks in five continents about there's a lot of interest. Yeah, there's a lot of interest. And I'm I'm in the middle of a.
00:21:02 Dr Peter Prevos
The series now it's it's fun to.
00:21:03 Dr Peter Prevos
Teach because yeah, for a lot of people, it's.
00:21:06 Dr Peter Prevos
Totally. They're entering in.
00:21:08 Dr Genevieve Hayes
So what's involved in your training courses?
00:21:12 Dr Peter Prevos
The book is really the is the course and how the book developed is. I started writing my own course notes and what I decided to do is very different to most coding books so this is not a coding book where you Start learning about a lot of abstract stuff in the first few chapters before you start writing something.
00:21:28 Dr Peter Prevos
Anything I got straight into it after just telling them about the basics of how to do arithmetic and.
00:21:33 Dr Peter Prevos
Those sort of things and what a console is and what a script is, I have a little case study about calculating the flow through a wire in a channel which.
00:21:40 Dr Peter Prevos
It's a very it's a first year civil engineering problem that most people that do the course from a theoretical perspective, but then for how do you solve this? How do you then write a function for this? And all the examples.
00:21:53 Dr Peter Prevos
Water based cause I'm I'm sick of analysing iris leaves and cars from the 1970s and those sort of data sets.
00:22:02 Dr Genevieve Hayes
Boston housing prices.
00:22:04 Dr Peter Prevos
Yeah. So it's all my data sets. They are synthetic because it is very hard to get datasets based on.
00:22:10 Dr Peter Prevos
My just pulling my next point and so they create all these datasets. I've got water quality data and it's real life stuff.
00:22:16 Dr Peter Prevos
So in the first course, for example, so which is, it starts from the absolute beginnings all the way to creating a automated PowerPoint presentation.
00:22:25 Dr Peter Prevos
That links and I did it that way because people who do the first course, at least they can see the full workflow.
00:22:30 Dr Peter Prevos
They're not to the left, hanging somewhere and teaching all means of functions, and I'm starting level 2 tomorrow and that goes into more on the statistics of doing and basis linear regression cluster analysis.
00:22:42 Dr Genevieve Hayes
One of the things I really liked about your book is that you take a very pragmatic approach to data science.
00:22:48 Dr Peter Prevos
That's that's exactly my point, and I that's how it evolved, because I'm I'm here to solve problems. I'm not here to.
00:22:55 Dr Peter Prevos
I'm not here to do research. Also, that's always been my approach as an engineer solve a problem with whatever data is available.
00:23:02 Dr Peter Prevos
And I know that's what my what my colleagues enjoy, but also talking to the publisher as I was writing it sort of spoke to the my contact over there. And he said well.
00:23:12 Dr Peter Prevos
So this is turning out to be quite a generic data science book. Could you market it wider? And he said no, we need to have industry specifics because there are so many data science books that are very generic and I think there's room, I'm pretty sure within other industries be specific about. I would love to read a book about agricultural science.
00:23:15
OK.
00:23:31 Dr Genevieve Hayes
What data science techniques do you find that keep coming up again and again in the water industry?
00:23:38 Dr Peter Prevos
Yeah, the book doesn't really or the course is really expressed that the techniques are that they're quite generic. The vast majority of the data in a in a water company or gas electricity, any sort of utilities times.
00:23:49 Dr Peter Prevos
Yeah. Whether that be customer compliance, whether that be pressure and water quality parameters and the hundreds of dollars, the all time series.
00:23:57 Dr Peter Prevos
So time series modelling is it's quite an essential skill which I haven't added to the calls because.
00:24:02 Dr Peter Prevos
It goes too deep.
00:24:04 Dr Peter Prevos
Things like cluster analysis don't happen often, and that's more in the marketing space. I'm working with customer team here.
00:24:12 Dr Peter Prevos
To look at.
00:24:13 Dr Peter Prevos
Better segmentation using some modelling techniques that we that need to be introduced a bit more are are normally detection because you know what utility, the ideal, what utility is invisible.
00:24:24 Dr Peter Prevos
Because do you know anything about you open a tap of water comes out. I I don't assume I'm. I'm not sure the technology in.
00:24:30 Dr Peter Prevos
There that you.
00:24:31 Dr Peter Prevos
Really care or or just in what?
00:24:32 Dr Peter Prevos
Happens beyond the tap.
00:24:34 Dr Genevieve Hayes
As long as the water comes out.
00:24:36 Dr Peter Prevos
And that's fine. But there's a massive world of complexity behind that which I love. I dwell on that, and what I'm proud of is that we can deliver you that experience, that you never notice anything.
00:24:47 Dr Peter Prevos
So normally detection is and I have a deal now with Latrobe University. We have selected a PhD candidate in Sri Lanka who's going through.
00:24:56 Dr Peter Prevos
The paperwork to.
00:24:57 Dr Peter Prevos
In the country we sponsor that person and that's a PhD in anomaly.
00:25:01 Dr Genevieve Hayes
Oh cool.
00:25:02 Dr Peter Prevos
Because anomalies are the most important, whether they are short-term predictive anomalies or so that we can manipulate our network to prevent a normally from happening.
00:25:11 Dr Peter Prevos
Also, and that's when you when you boil down to it, that's pretty much what we do looking at data and say.
00:25:16 Dr Peter Prevos
There's a spike. What do we do about that?
00:25:19 Dr Genevieve Hayes
Both time series analysis and anomaly detection are two topics that are not typically taught in your introduction to machine learning type course.
00:25:28 Dr Peter Prevos
Yeah. Interesting. Because I've never done any formal education in data science, so this is a whole chapter in the book.
00:25:34 Dr Peter Prevos
Or there's a substantial.
00:25:35 Dr Peter Prevos
Part of the book to talk.
00:25:36 Dr Peter Prevos
Because that's really what we do. You know, you measure a bunch of pressures and how often is it less than 20 water pressure, for example?
00:25:43 Dr Peter Prevos
That's those are the things that we live for as as water engineers, keeping that process taking over so that people can have their.
00:25:50 Dr Peter Prevos
Showers without thinking.
00:25:52 Dr Genevieve Hayes
Well, when I did my masters, I don't think we did any anomaly detection and time series analysis actually came up in machine learning for trading which was applying data science algorithms to financial data.
00:26:06 Dr Peter Prevos
So yeah, most of the stuff that I read is is about that sort of data. And and here's the other issue where where downtime and we deal with the real physical reality, we have sensors out there in the field.
00:26:17 Dr Peter Prevos
Who get hit by Lonos who are in?
00:26:20 Dr Peter Prevos
Frost or plus 40 degrees, they get soaked by water, so there's a lot of issues out there. It's quite a lot of work.
00:26:27 Dr Peter Prevos
To maintain all these sensors.
00:26:29 Dr Peter Prevos
So our data is not 100% reliable, but if I look at training data that is the data, there's no, there's no variance on that.
00:26:37 Dr Peter Prevos
If I, if I look at a a computer network, I can measure at 100% bid rates and whatever you guys look at.
00:26:44 Dr Peter Prevos
But this this is complicating aspect of looking at reality. So data quality is part of the nominee.
00:26:50 Dr Peter Prevos
So if we measure a pH of.
00:26:52 Dr Peter Prevos
Zero. Then, along with the probe.
00:26:54 Dr Peter Prevos
I think that there should be more attention to that, not just our industry and the industry that deals with physical assets probably where in in the book I talk about the DIKW pyramid, the data, information, wisdom pyramid that's been bandied around a lot. Well, I have my own version. I have removed wisdom because you study philosophy to.
00:27:06 Dr Genevieve Hayes
Oh yeah.
00:27:14 Dr Peter Prevos
Find that data science and I added reality at the bottom and that's what a lot of data scientists and analysts in general forget. Data science talks about a real real.
00:27:24 Dr Peter Prevos
Let's see.
00:27:26 Dr Peter Prevos
Even if talk about banking or Stock Exchange under the those numbers of the Stock Exchange is the reality of companies collapsing or being or redditors pumping up prices. So there's always a reality underneath the data and.
00:27:41 Dr Peter Prevos
As a pragmatic engineer.
00:27:44 Dr Peter Prevos
That's always what I look at. What I'm a really analyse.
00:27:46 Dr Peter Prevos
And my team, so I'm a civil engineer, I've build stuff all over the world. Pipeline St and one of my guys, Gary, he's a he honest he he has also has similar background.
00:27:56 Dr Peter Prevos
But I have two IT professionals who don't fully understand this. So we're now doing data excursions. Few weeks ago, we went to the reservoirs and we spoke to the reservoir keepers. What data do you collect? What does it look like?
00:28:07 Dr Peter Prevos
Reality. Where are these sensors? What? How do you measure this flow so that they actually see where the data comes from? And it's not just an abstract number on this?
00:28:16 Dr Genevieve Hayes
A couple of jobs ago this was about, I don't know, five or ten years ago I worked for worked for Victoria, which does the workers compensation insurance in Vic.
00:28:27 Dr Genevieve Hayes
I was the premium pricing manager there and if someone wrote in and had an inquiry about their premium or how their premium was calculated.
00:28:36 Dr Genevieve Hayes
I was the person who they got to speak to that person. So to write the letter.
00:28:41 Dr Genevieve Hayes
And I'd go out to meet with these customers, but it was only when I actually saw those customers face to face and heard about what the impact was of the premiums that I set, that I really understood that these weren't just numbers on a computer screen. These were actually numbers that would translate into money.
00:29:01 Dr Genevieve Hayes
That these people would have to pay and it could actually impact on the sustainability of a.
00:29:08 Dr Peter Prevos
That's a great example. Yeah. And that's robo depth is a perfect example of taking numbers in abstract reality without thinking about what it means.
00:29:16 Dr Genevieve Hayes
Yeah, I reckon if they had have taken the data scientists who developed that robo debt algorithm out into the field to speak to those people who are complaining, it would have been fixed a lot faster because suddenly it stops being a number. It starts being a person's face.
00:29:31 Dr Peter Prevos
Yeah. And it's work qualitative information. It's the quantitative and I can't keep stressing how important it is that the numbers can explain stuff, but only qualitative information is.
00:29:34 Dr Genevieve Hayes
Yeah, exactly.
00:29:44 Dr Genevieve Hayes
Yeah. And one of the things I love in your your first book, the principles of strategic data science are you give an example of all these different layouts of dots that all have the same mean and standard deviation and they're all completely different.
00:29:58 Dr Peter Prevos
You're my favourite data set, yeah.
00:29:59 Dr Genevieve Hayes
One of them's just used standard scatter plot of dots and one of them looks like a dinosaur.
00:30:04 Dr Peter Prevos
The data store is. I want a teacher. I want a teacher of the Datasource it's.
00:30:08 Dr Peter Prevos
One of my favourite visualisation.
00:30:10 Dr Genevieve Hayes
I'm teaching data science next semester. That visualisation is going to appear on my slides.
00:30:14 Dr Peter Prevos
Right. Yeah. It's a great paper, really. The original. There's quite a bit of.
00:30:18 Dr Genevieve Hayes
So how many times have you delivered your data science Unicorn pro?
00:30:21 Dr Genevieve Hayes
OK.
00:30:22 Dr Peter Prevos
Close to a dozen I think I have a record somewhere. They keep coming back there. There's people on the wait list at the moment.
00:30:28 Dr Peter Prevos
So we're talking about the next session. I'm, I'm I'm sort of ready for the next thing really, but it's not turned into the book I'm planning.
00:30:35 Dr Peter Prevos
To do the video content of the course and put, just put it on our website to deliver it all the time.
00:30:41 Dr Peter Prevos
I got bored with my own challenges and then I need to move something different, so I'll put that on my website.
00:30:46 Dr Peter Prevos
Yeah, maybe, maybe also a way to sell the book, but just put the screencasts on our website with a little water problems.
00:30:52 Dr Peter Prevos
And what I'm interested in now and and also other people in my industry, there's more more data scientist is there are not enough datasets like example problems for our industry.
00:31:01 Dr Peter Prevos
To play with.
00:31:02 Dr Peter Prevos
So I had. I had to synthesise data, but now I found there are some datasets, but Kaggle or and other machine learning data repositories have not. There's almost nothing there.
00:31:12 Dr Peter Prevos
About water or water quality or water resource, which is a really important problem because the world is running out of water or we're using too much.
00:31:19 Dr Peter Prevos
The way you.
00:31:19 Dr Peter Prevos
Want to?
00:31:20 Dr Peter Prevos
So I'd love to get together and I've got lots of international, so hopefully we can set up a repository for water data and get non water experts interested in maybe competition. So that's I think that's my next.
00:31:32 Dr Genevieve Hayes
Us interviewing two women who are involved in environmental science a few episodes ago and when I mentioned to them that I was interviewing a man from the water industry, their first question was when is this interview and how can they get?
00:31:47 Dr Genevieve Hayes
How can they get hold of it? Because so there are people out there who are very interested in water and I'd say.
00:31:53 Dr Genevieve Hayes
Given the high interest a lot of people have in environmental sciences nowadays.
00:31:58 Dr Genevieve Hayes
Water is very close to that.
00:32:01 Dr Peter Prevos
Obviously plays a big role in the environment and in in our own life, and we work very closely with environmental science.
00:32:06 Dr Peter Prevos
Unfortunately, I didn't.
00:32:07 Dr Peter Prevos
Have sort of room.
00:32:08 Dr Peter Prevos
To to management case study and.
00:32:10 Dr Peter Prevos
We do work with there are.
00:32:12 Dr Peter Prevos
Not for profit groups who come to us and ask.
00:32:15 Dr Peter Prevos
For data about.
00:32:16 Dr Peter Prevos
For example, we have excellent data about PET.
00:32:18 Dr Peter Prevos
Sites and so we can freely share that, but it wouldn't be lovely if we had a data repository for we're all what easily just freely provide that data there.
00:32:27 Dr Peter Prevos
There are no secrets as far as I'm concerned. There's a little bit of that's where groups like those environmental groups.
00:32:34 Dr Peter Prevos
Can get their data where.
00:32:36 Dr Peter Prevos
People who are interested and one of the developer.
00:32:37 Dr Peter Prevos
Products can get.
00:32:38 Dr Peter Prevos
You can organise a hackathon, so that's that's. That's my. That's on my wish list for like the other case studies get people to share more and yeah, see, see how we can develop that further.
00:32:50 Dr Genevieve Hayes
I could also imagine this would be of great value if you're considering.
00:32:54 Dr Genevieve Hayes
Water data sets from developing countries where they don't have clear water coming out of their taps.
00:33:00 Dr Peter Prevos
Yeah, that's that's a a great challenge. If those data sets are available, perhaps entrepreneurs or startups can we've we've done some work with, with my organisation column and water in in Vietnam to provide drinkable water out of it.
00:33:14 Dr Peter Prevos
There's a lot of interest and some of my colleagues went to Ghana. There's.
00:33:17 Dr Peter Prevos
A lot of work happening.
00:33:19 Dr Peter Prevos
And and perhaps the data.
00:33:20 Dr Peter Prevos
Can playing a role of that tool to help these people to get better?
00:33:23 Dr Peter Prevos
At analysing their own.
00:33:24 Dr Genevieve Hayes
Data science for social good type thing.
00:33:26 Dr Peter Prevos
And what is actually amazing?
00:33:28 Dr Peter Prevos
A lot of people, really.
00:33:28 Dr Peter Prevos
I said Africa, we have this idea of Africa. Still. It's a dark continent, but mobile technology they have actually leapfrogged us.
00:33:36 Dr Peter Prevos
So they went straight to mobile technology. Mobile payments are are quite normal there. I've seen some presentations from African reward utility, so really advanced.
00:33:46 Dr Genevieve Hayes
Legend has it that unicorns possess magical powers, so I think it's very appropriate that you 2 have an interest in theatrical magic.
00:33:55 Dr Genevieve Hayes
In fact, in addition to writing about data science, you're also the author of the book Perspectives on Magic, Scientific views of Theatrical Magic.
00:34:05 Dr Genevieve Hayes
I think this is incredibly exciting. When I was a kid, my dream career was to be a stage magician, so I used to do magic shows for my parents and family members and anyone who I could get to sit down in front of me for half an hour or more. And when I was in grade six, I remember reading about a school for.
00:34:26 Dr Genevieve Hayes
Professional magicians that had opened in Melbourne and that was my tertiary education plan.
00:34:33 Dr Genevieve Hayes
Can you tell us a bit about your interest in?
00:34:36 Dr Peter Prevos
Well, my back story is just like yours, so I used to annoy all my friends. Even started doing some performances for external people and I I was quite serious about it, but my interest in magic sort of waxed and waned through my life course, as magicians call it, getting distracted by life. So I went back for magic because I sort of saw it as a bit of a futile.
00:34:57 Dr Peter Prevos
Cheesy exercise Cause magic has that risk of being cheesy, but then it's about two or six. I think somebody asked me. My in-laws asked me, can you do a magic show for the kids?
00:35:08 Dr Peter Prevos
I thought, Oh well, let's let's try to sing out again and develop a.
00:35:11 Dr Peter Prevos
Full 45 minute magic show for kids.
00:35:14 Dr Peter Prevos
It still has my interest and when I was doing literature research for my dissertation, looking at training management training, I found this paper about using magic tricks or and I totally got distracted and during my PhD I wrote.
00:35:28 Dr Peter Prevos
This book of.
00:35:28 Dr Peter Prevos
Scientific views of on magic cause the more I started looking and having access, I found hundreds of papers about lot, most of them psychology about how you know why I can we be the history of magic, the anthropology of magic interact with each other. Perspectives from from theatre studies, obviously.
00:35:46 Dr Peter Prevos
So the book is pretty much a an annotated bibliography and computer science plays a role in that. It's a few papers written about that because a lot of magic tricks use mathematics to to actually create the illusion of magic.
00:35:59 Dr Peter Prevos
What magic really is is that there's there's there's two things happening. There's whatever the processes that you see and there's a process that's hidden.
00:36:06 Dr Peter Prevos
From you there are interesting bits of magic number theory and also geometry and topology that are counter intuitive. And when you use those then you create a magical outcome with a bit of presentation skills and there is a really interesting website, computer science for fun C4F N.
00:36:22 Dr Peter Prevos
Org that this is the whole website dedicated to computer science and a really interesting paper from early 90s I think I can't pronounce the guy's name, but it was about how I mentioned earlier how computer interface is really imagined.
00:36:34 Dr Peter Prevos
It's an illusion that you think you're dragging a.
00:36:36 Dr Peter Prevos
An actual physical.
00:36:37 Dr Peter Prevos
Folder from one file from one folder to another and he talked about how.
00:36:42 Dr Peter Prevos
Perhaps if computer interface designers can learn from magicians and how to make those?
00:36:48 Dr Peter Prevos
And there are there are magicians such as Marco Tempest from Switzerland, and look him up on YouTube. His stuff is fun because it's it's high tech net with with iPhones and all sorts of things.
00:36:58 Dr Peter Prevos
So there's quite a bit of overlap in computer science, and my interest at the moment is is mainly index and topology and optical illusions. Running a book about I've written some E books about.
00:37:09 Dr Peter Prevos
Certain magic tricks that that have been around for a while on this historic research and work topologically. It's just just fun to.
00:37:16 Dr Peter Prevos
To play with.
00:37:17 Dr Peter Prevos
With these concepts.
00:37:18 Dr Genevieve Hayes
Few minutes ago you said something about using magic tricks for management.
00:37:24 Dr Genevieve Hayes
How does that work?
00:37:24 Dr Peter Prevos
Yes. Well, just a great team building thing to get a group of people together and teach them how to cut a rope and then restore it again and then and then with an analogy behind it, you can you can make it up as you go along almost a lot of magicians actually make good money performing at corporate events, a little act that I'm working on is magic and innovation.
00:37:44 Dr Peter Prevos
It's really fascinating for me to read. I'm all not not like a magic scholar than a performing magician is.
00:37:50 Dr Peter Prevos
I'm magicians actually innovate some. Some some mathematician comes up with a print edition says, oh, I can use that.
00:37:55 Dr Peter Prevos
And then you can see that because all the literature is now scan them online and your business is going to learn from the whole magician innovate.
00:38:02 Dr Peter Prevos
To be a good innovator, you have to be willing to fail. You have to be willing to flatten your face. Now every magician and you will have that experience has died on stage.
00:38:10 Dr Peter Prevos
So. So right, so so here's an analogy. How you can tell a magic story. Do a magic trick with it and say, well, if you want to be a good innovator, be willing to die and and and make a fool of yourself. So that's those sort of things I think, are the the lessons that the world.
00:38:24 Dr Peter Prevos
Apps can load.
00:38:25 Dr Genevieve Hayes
When I was doing my PhD, I lectured for several years and now I'm back at it, and that was, I think, one of the best skills I learned.
00:38:33 Dr Genevieve Hayes
You inevitably die on stage and someone will point out, hey, you just messed up that problem that you did on the whiteboard.
00:38:41 Dr Genevieve Hayes
Getting through that without bursting into tears or running off in embarrassment, valuable life skill.
00:38:48 Dr Peter Prevos
Exactly, and that's that's anyone who's done magic will have great presentation because being able to make the best out of it themselves.
00:38:55 Dr Peter Prevos
So it's it's just.
00:38:56 Dr Peter Prevos
A pile of my life. I'm going to Las Vegas to do a one week intensive course for it's for professional magicians, but yeah, it's just fun fun to be. And among other magicians. And see.
00:39:07 Dr Peter Prevos
How they think?
00:39:07 Dr Peter Prevos
And they have totally different lives.
00:39:09 Dr Peter Prevos
Than we have and it's just this little hidden world that that's fun to be part.
00:39:14 Dr Genevieve Hayes
I'd like to pick up on your comment about mathematicians as innovators because it actually reminds me of something that a previous guest was telling me about how his company would often take academic research and then innovate on that produce a commercial product. If you're a data scientist who was wanting to.
00:39:34 Dr Genevieve Hayes
Learn innovation skills from a magician. What would be the first steps to take in doing?
00:39:41 Dr Peter Prevos
When you look at how how magic tricks are created, a lot of magicians create their own stuff, and their creativity is just like any other art form really is.
00:39:49 Dr Peter Prevos
It starts with playing goal, is playing muck around with props for data science that will be marked with some code but not have a purpose in mind. Also put some time aside when I was in a previous job.
00:40:02 Dr Peter Prevos
I always carved out to time. I still try to do that Friday afternoons. It's gonna play and say, oh, what about what if I visualise this data this way? Is that is that is that useful for and then half the time nothing comes out of it.
00:40:15 Dr Peter Prevos
But yeah, sometimes some really good stuff comes out of that play and and some of the best things that are developed are and other business for more than 10 years came up, came about from there.
00:40:25 Dr Peter Prevos
And as we said earlier, not not being afraid of failure, so goaless play, but also when you design design A magic actor, I'm not a magic designer, but.
00:40:35 Dr Peter Prevos
The people who.
00:40:36 Dr Peter Prevos
Well, really good at it. And Jay Sankey from cancer is an amazingly creative magician, and he talks about.
00:40:41 Dr Peter Prevos
Don't worry, don't worry about the methods set out. The goal that you want to.
00:40:45 Dr Peter Prevos
And then and.
00:40:46 Dr Peter Prevos
Then sort of went backwards on what of this that you, let's say my goal would be want to have an algorithm that solves.
00:40:52 Dr Peter Prevos
A certain problem.
00:40:52 Dr Peter Prevos
Then and set that goal and then work out what tools and and just keep chipping away at it. But yeah, failure is always an option.
00:40:59 Dr Genevieve Hayes
Is there anything on your radar in the AI data and analytics space that you think is going to become important in the next three to five years?
00:41:09 Dr Peter Prevos
Everybody's talking about generative AI, of course. The, the, the topic of the day I think is going to be important in in different ways.
00:41:15 Dr Peter Prevos
But what I think generative AI can be very good at is solving this problem within my organisation. We have really good data generated, the massive suite of insightful reports. Most of my colleagues aren't even out of they've forgotten about it or wasn't intended for them in the first place.
00:41:30 Dr Peter Prevos
So here.
00:41:31 Dr Peter Prevos
My vision that we are probably our users, what I want is to go into power BI and you say which town had the highest number of water quality complaints last financially bang and graph shows up.
00:41:42 Dr Peter Prevos
That's and and I know not exactly, but I think how it would work, but I could sort of envisage that it's within the realm of possibility with a lot of.
00:41:51 Dr Peter Prevos
Cause we have so much data now. So many reports nobody has time to look.
00:41:55 Dr Peter Prevos
And that's and that's also part of the PhD research that we're starting.
00:41:59 Dr Peter Prevos
Anomaly detection as a method to sort of wade through.
00:42:03 Dr Peter Prevos
The multitude of reports, so that's my colleagues don't have to. So we have 19 water systems.
00:42:08 Dr Peter Prevos
In what plant? So for now them to monitor that they have to look at reports and graphs and, but if we have an anomaly detection tool that can be smart about it, we can display anomalies to them and forget about the rest.
00:42:20 Dr Genevieve Hayes
I think that would be useful not just for data, but for knowledge in general because.
00:42:25 Dr Genevieve Hayes
If I think of some of the larger organisations I've worked at in the past, most organisations now have something like a SharePoint repository or some sort of knowledge management system and you can never find anything in it.
00:42:37 Dr Peter Prevos
Yes. Yeah, my my manager. So one chibita became popular. We did a little competition. I asked Chibiki water management question and our executives and we had a scoring system. The executive still won.
00:42:50 Dr Peter Prevos
So the humans, but only just. So then the challenge was put to me. Oh, can we just run this language model over all our garbage?
00:42:57 Dr Peter Prevos
Really, on our sheer drives and and so well potentially we could if if you have spending our computing power, you could train a mobile on all the stuff you have sitting on your. So I'll start a toying with fine tuning GT but.
00:43:09 Dr Peter Prevos
Not enough time.
00:43:11 Dr Peter Prevos
To do that, I'm not sure if it's that's useful yet, but.
00:43:15 Dr Peter Prevos
I could see it moving there.
00:43:16 Dr Genevieve Hayes
I think Microsoft has a plan to make that possible for everyone going forward.
00:43:21 Dr Peter Prevos
Yeah. Anything that's stolen is.
00:43:23 Dr Genevieve Hayes
And what final advice would you give to data scientists looking to create business value from data?
00:43:28 Dr Peter Prevos
Yeah, start with a real life problem. If you're interested in a certain industry, let's say environmental management. There's lots of problems out there that you could be, but most importantly and we mentioned before is to understand the physical reality or the social reality, what it is that you do with go go beyond the numbers, go out in the field and maybe even take some samples yourself.
00:43:49 Dr Genevieve Hayes
For Looseners who want to learn more about you or get in contact, what can they do?
00:43:54 Dr Peter Prevos
Jump on LinkedIn. I'm on there. If you Google Peter Prevos and you probably get very close to my website lucidmanager.org or Peter prevos.com as my landing page.
00:44:05 Dr Peter Prevos
And on the Lucid manager website, whenever I feel like writing stuff on there, so my course etc will be on that.
00:44:12 Dr Genevieve Hayes
And I can put a link to that.
00:44:13 Dr Genevieve Hayes
In the show.
00:44:14 Dr Genevieve Hayes
Notes. Thank you very much for joining me today.
00:44:14 Dr Peter Prevos
That'd be nice.
00:44:17 Dr Peter Prevos
That was fun, Jenny.
00:44:19 Dr Genevieve Hayes
And for those in the audience, thank you for listening. I'm doctor Genevieve Hayes, and this has been value driven data science brought to you by Genevieve Hayes Consulting.
Creators and Guests
