D (2:58)

Hi Dervika, first of all, thank you very much for having me. And also greetings to everybody who is going to listen to this episode wherever and whenever, in the future and all around the globe. So yeah, now we are talking about a book, but actually it didn't start as a book. So the beginning was really humble and modest. So I just wanted to write or curate a simple cheat sheet where I wanted to put all my essential and frequently used Python snippets, which I used for geospatial data science. So I had loads of sticky notes where I had a few lines of code, for example, to how to read certain datasets properly or how to fine tune certain map visuals. And I realized that it would be probably helpful in the future for me for sure, and maybe even for others, if I somehow compiled this into a sensible cheat sheet. So I started putting together all these items I had scattered all around the Python notebooks and endnotes. And then I realized that it's gonna be quite a lot. And somehow this idea of having 101 so like a 101 series stuck with me. So then it wasn't too difficult to like make the list grow a little to reach that mighty 101 number. And then I realized that this is actually a lot more than a cheat sheet. Maybe like 10 different cheat sheets, but also it was purely technical, so there were a few titles and headlines, but it was essentially loads of Python code. So I decided to add a few explaining sentences on what this or that piece of code does. And once I have finished that, which took probably a few weeks or even a few months, I took a step back and realized that, hey, here I've chapters, I have sections, subsections, code visualizations, and quite some text. So this actually looks very much like a book. So before sharing it everywhere, I decided to repackage it and rebrand it and call it a book. And then of course there were some technical parts which we will discuss later. But then it eventually quite naturally evolved into a book format from a simple cheat sheet. So, long story short, it was not planned, it was merely an accident that.

C (5:19)

I wrote this book, this is such an interesting story. So when you were trying to think about that this cheat sheet is now something which can be turned into a book, did you have a target audience in mind that who this book could cater to?

D (5:33)

Well, when I started to put together the cheat sheet it was not very clear, but at later stages when it started to take shape, I realized that this is sort of a material which I would have loved to have when I started to work with geospatial data. So also fun fact, I had no idea I was doing spatial data for at least a year when I was already doing it. So I started completely random picking up very random scpaths from Stack Overflow. Yeah, that was the before or pre chatgpt era. So I thought that this list should be something which I would have loved to see five years ago when I started doing Python and geospatial together. So to answer the question very quickly, the audience was whoever wants to learn geospatial data science in Python, or want to upgrade their skills or like fill all the gaps at the foundational parts of geospatial data science in Python?

C (6:31)

Very interesting. I wish I could have something like this when I started doing the spatial data science, but better late than never, right? Yeah. So this is, from what I understood that you tried to put together a lot of Python tricks and tips which you were using in your work. So could you describe a little bit about the content of the book, what the book covers, what it doesn't cover?

D (6:56)

Yep, sure. Actually there was even a data driven part of the decision making process on what to put in there. So I wrote a short Python script to parse hundreds or even thousands of my Python notebooks and I simply did a frequency distribution of the most widely used packages and most widely used comments. So I kind of had a list of the most frequent things which was already quite good and straightforward to decide and draw up the main parts of the book. So that's how it starts with vector data, then it dives into different topics like roster data or OpenStreetMap, which is of course as we know, here in the geographic field is a very widely used data source.

C (7:43)

Great. So one of the things I found very interesting about the book is that unlike many technical book, your book emphasizes on tips and tricks rather than going into deep advanced theory about spatial data science. So why did you choose this practical first approach and what advantage does it give to beginners to read this book before they deep dive into the field of spatial data science?

D (8:08)

Well, on the one hand I have Proper academic background. So I studied physics, biophysics, I did a PhD. So I deeply value deep academic knowledge. But as I have been in the industry for about five years now, I also realized that it's very rare in practical settings when people have the time and the bandwidth to dive into the theoretical part. So what we usually need in practice, which is, which might be very superficial, but very often it's the reality. So that's what I think we need to adapt to, is something quick which is like, you know, get straight to the point and assume that all the theory is right in the background because, well, it's probably write if there are books on those theories. But at the end, if we just want to get a project done quickly, then we will also need to be able to learn quickly and fast. So that was kind of the motivation. There are great books out there with proper theory, so I thought that, okay, that's there. Now let's write something which is really straight to the point and like a quick shortcut to get there.

C (9:14)

Very interesting. And I think this book fills a sort of gap in the literature in that sense where you get the guide instead of getting a full whole on theory, an advanced theory book. So as you said, it might be a right starting point for people entering the field of spatial data science. And I found that the topic of the book is very relevant and timely. Spatial data science is emerging. It is growing as a field. Why do you think that it is the right time to learn spatial data science and to learn the foundations which are provided by this book?

D (9:48)

The first time I've worked with spatial data in Python, so in data science settings was in 2018. And if I look back to the seven years then I clearly see accelerating trend during the past maybe two years when spatial data in general is getting available at a much higher scale and a much better quality. And also the tools and the development of tools are, are speeding up very rapidly. So for instance, now of course we are in this AI era where new models are being published every week, but just during the past month there was at least one major GeoAI announcement per week. So all the big companies were launching their new GEO AI platforms. So that's. And for example, five years ago, I don't know, maybe it was months without such big announcements at all. Or another example which I like a lot about a very simple computation on spatial graphs on road networks, which I think six years ago took maybe two or three weeks on a large cluster and now it runs in like 15 minutes on a MacBook Pro. So things are speeding up and of course AI is skyrocketing. But I believe that after the AI hype, geospatial is one of the fastest growing parts of data science.

C (11:12)

Totally. I totally agree to that. And while reading the book, I was very intrigued by the content that you cover, which is very basic, but which is very important for a beginner to know for or for anybody who is working in spatial data science to know and understand. So I wanted to discuss more on the content of the book for our audience to see what they can expect in terms of what the book covers. And I found this very interesting chapter that you talk about Python Stack. You cover a lot of rich toolkit libraries, Shapley, Geopandas, Folium. Which of these do you think are must know tools for somebody just entering the field and how they can pick up on these?

D (11:56)

It's most certainly Geopandas. So anyone who has already been working in Python and did at least like one hour of data science in Python must have met Pandas, which is for those who haven't done that, one hour of Python DSIS yet is essentially like an Excel on steroids in Python. So it's designed to manipulate and deal with tabular data of any sort. So anything that can be put in a spreadsheet is good to go with Pandas in Python. And Geopandas is essentially extending Pandas into the geographic field. So there every record, which still we can imagine as a row of data in a spreadsheet is also assigned to a specific geometry. This geometry can be a location on a map, it can be the boundaries of a country, or even it can be, for example, a route Google Maps is planning for us when we go out to grab a coffee, where the coffee shop is marked by a pin. So pair of coordinates, which is again a geometry. Yeah. So Geopandas is like, I think that the Swiss knife of spatial data science. Of course it also comes with limitations when the data is too large or too complicated. But in my experience the 8020 rule holds very well. So like 80% or maybe 95% of the datasets or vector data can be very well handled with geopandas.

C (13:30)

Yes, totally agree to that. And that's one tool I have been using a lot in my work too. So I see that you cover a lot of good examples of these Python toolkits in the book from which the users can start reading and grasping on the concept of this. Could you talk a little bit about what kind of examples are covered for the Python toolkit?

D (13:49)

Yep. So of Course, we are covering Geopandas, which is designed to deal with vector data. And in later chapters we are also covering its counterpart called Rasterio, which is to handle raster data. So these are the two main libraries, I would say. We also cover a couple of other tools. For example, Shapley, which is fairly rarely used directly, but it's everywhere. So Shapely is a library which creates geometries. Probably it's not in the book, but I like to play this exercise when I teach these materials is to use Shapely to draw different figures because, you know, it's geographic, but it's also coordinates. So at the end, it doesn't matter if we are drawing a, I don't know, a smiling face or we are drawing the boundaries of the United States. At the end, it's a set of coordinates. So it's very low tech. But I think that it helps a lot to understand how the spatial dimension of these datasets actually work.

C (14:57)

Yes, that's a very good point that you raised the spatial dimension, the coordinates. And I see that in the beginning of the book you cover a lot of these fundamental topics, geometries, coordinate system and map projections. Why do you think mastering these fundamentals are so important before moving to advanced spatial data science?

D (15:18)

That's mostly coming from my own learning from my own mistakes. So first the geometry part was almost a complete black box to me. So it was just one column in a large spreadsheet. But it's usually so much more than just a spreadsheet with like floating point or integer numbers. So that's why I decided to like crack it down and rip it into pieces. And even if it's slow and very low tech and maybe even boring, I think it's really worth spending two hours on reviewing it from scratch. And the other part is that especially if we are iterating between platforms, GIS platforms and Python, we are working with older data, data sourced directly from any kind of like measuring device, then usually geometry part is where things can easily go wrong. For example, if we don't have the right projection system, then actually we cannot even be sure that we can use the data because then even if we have the data, we don't know how to map it to any reference coordinate system, then the whole thing will be just garbage. That's usually so on the one hand it's simple because it's just a series of coordinate pairs, but on the other hand, since it's not just one integer number, but it's many floating point numbers, well, the complexity can blow up and that's where things can easily grow go wrong. So to cut it short, it's very helpful when debugging random data sets coming from all sorts of like either repositories when we know the geometry is inside out.

E (16:58)

When did making plans get this complicated? It's time to streamline with WhatsApp, the secure messaging app that brings the whole group together. Use polls to settle dinner plans, send event invites and pin messages so no one forgets mom 60th and never miss a meme or milestone. All protected with end to end encryption. It's time for WhatsApp message privately with everyone.

F (17:22)

Learn more@WhatsApp.com Mint is still $15 a month for premium wireless and if you haven't made the switch yet, here are 15 reasons why you should 1. It's $15 a month.

C (17:34)

2.

F (17:34)

Seriously is $15 a month.

C (17:37)

3.

F (17:37)

No big contracts.

D (17:38)

4.

F (17:39)

I use it.

A (17:39)

5.

F (17:40)

My mom uses it.

B (17:41)

Are you.

F (17:41)

Are you playing me off? That's what's happening, right? Okay, give it a try. @mintmobile.com Switch upfront payment of $45 per.

C (17:48)

Three month plan $15 per month equivalent required New customer offer first three months only, then full price plan options available, taxes and fees extra. See mintmobile.com that's very true. And mostly we learn it the hard way. So it's better to learn it through the book than link it by. Make it better, right?

D (18:03)

Yeah. But yeah, hands up those who haven't done any map which looked like insanely distorted and had the most random shape ever.

C (18:11)

Yes, I think most of us started there and that's when we realized the importance of learning the fundamentals. Well, you dedicate one full chapter on full chapters to raster and vector data. And what specific skills and concepts to do you want the readers to gain from these chapters? And how do you think these two chapters complement each other in terms of skill the readers would learn from them?

D (18:40)

Maybe we should first set the definitions clear to make sure that we are all on the same page. So vector data, if we just want to take a few examples, is information which can be described by, well, directional vectors if we stick to the coordinate geometry concept. So points, lines and polygons where no matter how much we zoom in, we will still see the same shape. But in rasters we have discrete grids with a given number of pixels. So while vector data is abstract vectors on a coordinate space, raster data is actually like a picture, an image where we can zoom in, but the more we zoom in, the more pixelated the Image gets and usually data comes either in one structure or the other and learning to combine them because we cannot just live our life only working with vector data or only working with raster data. It's very important to learn how to read these, how to process these, how to clean these, how to set their projection systems right, how to visualize them. And if we learn the basics of that, then probably we will be very relaxed regardless of what kind of data we will come across.

C (19:54)

Right? Yeah. So one very interesting thing I found in the book that especially since open source GIS tools and data sets are growing and I see you dedicate a whole section to open street maps, Right. How do you see the role of open and crowdsourced data shaping the future of geospatial Data science?

D (20:20)

Yep, OpenStreetMap. I think that I'm not alone whose probably favorite go to data source is awesome because it's just so rich. There are so many different things. It has such a wide coverage and of course it's easily and freely accessible. Whatever data project I work on, OSM is one of the first sources I check out and I think that we are moving to a direction where open source spatial data will be well now it's probably not superior in terms of quality compared to propriety data providers, but because especially that the field is growing so fast, there are so many more new tools and also new people contributing to these that probably sooner than later open source geospatial data will be just as good in quality as propriety data without all those restrictions which cover commercial legal restrictions and also like the number of API requests we can send. So this I believe slowly pushing us towards a world where we have like a complete digital twin of the whole planet in decent like vector and raster formats. Also something I see as a new trend is that many of the large global cities have launched their own GIS portals, many of them already doing lidar scans and building 3D models of the cities. So on the localized level also really high quality data is being produced as we speak. And once these really high quality local data will meet large open platforms like osm, that's going to be a very fabulous moment for geospatial researchers.

C (22:00)

Great. And in terms of the examples that you cover for OpenStreetMap, are they part of the projects you have been doing or what can the readers expect in terms of what kind of examples they can see in the book?

D (22:15)

Yeah, probably there it is a bit biased because I've been working with a lot of urban planning projects lately and also I did my PhD in network and data science, so spatial networks are also pretty close to my heart. That's also the reason I dedicated a whole chapter to it. So maybe that part is not that essential for anyone who is jumping into spatial data science. But. Well, this started as my like private cheat sheet, so I wanted to make sure it's there and also spreading the world of how, how cool networks are and of course how useful they are.

C (22:52)

Yes, since you touch on that topic, I wanted to cover that later, but let's talk about it now. Towards the end you talk about advanced topics like spatial networks, machine learning. So the users do get a glimpse of advanced topic. So could you talk a little bit about that? What kind of examples are covered in terms of these advanced topics? What can the reader expect to what depth do you go in the advanced topics?

D (23:20)

Yeah, sure. So, well, this is an intro book, so even if the topics are a bit more advanced than the very basics, I wanted to make sure that all the codes are easy to read, easy to interpret, and people can just use them in a grip and go away. So there are no complex dependencies. And it's quite complex, at least I hope it is. And for example, for networks, a great example is accessibility. So accessibility in the spatial data context and in urban settings means computing how easy it is to reach different parts of an urban area. So to be a bit more practical, let's assume that we live in a given neighborhood and there are a couple of doctors offices in that neighborhood. And using spatial networks, we can compute for every single point in our neighborhood, how many minutes does it take to reach, for example by walk the nearest doctor's office? And such computations are actually meeting a quite famous urban planning term or concept very well, which is called 15 minute cities. So 15 minute cities means that a city has a high quality of life. If all the necessary services for our daily lives, like doctor's office, our workplace, kindergarten, grocery shopping, and all these things can be reached within 15 minutes using either public transportation or walking or biking. If I remember correctly, public transport is also part of it. But maybe in the comments I will get corrected Anyway, so it's a very nice concept, but how do we validate or how do we test if a certain area satisfies the criteria of a 15 minute CD? And the answer is spatial data science and spatial networks. So. So that's also a good reason spatial networks are part of the book.

C (25:19)

I totally agree to that. Spatial networks I have been using a lot also for healthcare accessibility analysis, so I think it's a very important concept to know about and learn before deep diving into advanced spatial data science. Well, you said it's a prep and go book. That's the line which caught my attention. So who do you advise? Are the people who can just grab the book? Maybe somebody going for a geospatial data science interview should just go through quickly and then. Or what, what, what do you. Can you elaborate a little bit more on the prep and go part?

D (25:57)

Yep, yep. So ideally, and what my intention was that anyone working with this geospatial data set or. Oh yeah, let's roll back a little. So let's assume or like do some little thought experiment, someone who hasn't done spatial data science yet gets an email from a colleague that, hey, there is this data set which has coordinates in it. So how can we make some sense of it? And once we receive that email and that dataset, probably in a spreadsheet or something even more mysterious like a geojson or a shape file, then we can just grab this book and start figuring out how we want to analyze the data. Because this book, this book will help to read the data, turn it into a data frame, explore the basic features, create visualizations, and then depending on the spatial features it has, this book will allow you to dive deeper into, for example, the network component, if there is any, do some geocoding, if some addresses or locations are missing. If it's too large, then you can also try methods from spatial indexing. And of course we can also get to the machine learning part and do some regression modeling or even some simple forecasts.

C (27:14)

This reminds me of what you said at the beginning of the interview, that sometimes we are doing spatial data science without knowing we are doing spatial data science. And that is something I have heard from many people I have worked with who are doing spatial data science and they are not aware of it. So that makes me realize that though we are pitching this for spatial data scientists, but actually after reading the book, I found that it could be for anybody working with spatial data. And if, like you said, if you get a sheet with coordinates, JSON, shapefile, then this book could be the right one for you to read before even trying to explore that further. Do you have any comments on that?

D (28:01)

Yeah, so probably this book is not for spatial data scientists, but for data scientists working with spatial data.

C (28:08)

Yes, totally agree to that framing of the. And another very interesting thing I liked actually about the book was as a spatial data scientist or as data scientists working with spatial. Like you said, we focus a lot on the analysis part, but there is sort of the visualization part sort of gets less attention, which I think is equally important because ultimately at the end the goal is to produce cool good looking maps to convey your results or ideas in terms of visualization rather than big sheets. Right. As geographers we want to get it out as a visual as a map. But your book dedicates a lot of space to visualizing data with Matlab plotly Pydec. So could you talk a little bit more about what kind of visualization tools are covered in the book and in your own work? How do you decide the best visualization tool and so on?

D (29:09)

Sure. Before that, a quick comment here, which I have noticed and might even be an unpopular opinion, but it's more like an observation. So I have been running a one day one paper day post series on LinkedIn from the first day of 2025. Now we are at like day 250. And what I noticed is that on the one hand academic papers are usually not done by designers. So the maps like get the job done. They have all the information, all the labeling is right. But not all the maps are as pretty. And when I say pretty I mean it the most like layman term possible. So it's not an eye catching visual. It's a academic piece but still might be a little less pleasing than a fancy infographics. And what I see is that the nicer looking a map is, the more likes and recognition it gets on social media. So even though on the one hand people writing the papers don't seem to be interested in that and is probably judged as being superficial because the science and the content should be what matters. But at the end it seems that people who are reading it will stick to the ones with the nicest visuals. So it's more like a gateway to the reader's attention. Especially in today's world when we have so much content in our feed, it seems to me that having the right visual language for even the most academic paper is probably a must just because now we need to be seen because there is just so much stuff out there competing for our attention.

C (30:53)

So I had I actually there were a lot of takeaways for me after reading the book, but my main takeaway was like get your fundamentals right before even diving into Geo AI, Geo LLMs, Gen AI and spatial data science. It's so important to just and even just like brush up your fundamentals every now and then. This book would be a great sort of revision guide. Anybody wanting to brush up their Skills. But I wanted to know from you if there is one takeaway key takeaway you would want the readers to learn from the book, then what would it be from an author's perspective?

D (31:31)

That's. That's a good question because. Well, you know, it started as 101 takeaways, but probably it's something like get your data. Right? So if you have a special data set, then we should be able to read it and we should be able to describe all the core features. So what's the crs, what's the spatial coverage? Whether the geometry component is accurate and all the geometries are valid. So the ability to be able to find, recognize and get a spatial data set ready, and probably that rhymes well with the garbage in, garbage out phrase or collective wisdom. So I think that here, probably the most important part is to learn how to make sure that the spatial data set we have is in a good shape.

C (32:20)

Right. The spatial data is in the good shape, then the results and the findings will be in a better shape. Right.

D (32:26)

And a quick map visual is usually a very good way to tell how much of a good shape the data is.

C (32:33)

So. True. Well, Amilan, I know that you have been working on interesting stuff. You always have some interesting stuff coming up for spatial data science on spatial could. Some of it I'm aware of, but I would love to hear from you on what next, what the readers can expect next coming from you, if you can talk a little bit about your future projects.

D (32:55)

Sure, gladly. Well, there are a lot of projects, but probably the one that keeps me the most excited these days is my course series on LinkedIn learning. So. And I also have a few courses on Udemy and I just started to put them together in like a learning journey. So now it's not just a couple of courses on this or that topic, but now it is turning into like a proper, I would almost say mini program. Now it is getting to have a really nice arc focusing on different topics in improper online courses. So if someone wants to keep hearing me more about talking about Python instead of like reading the book, then that course is going to be the right place to follow up.

C (33:41)

Oh, that's very interesting. So do you want people to read the book first and then go to the course or do the course first and then come back to the book?

D (33:49)

Probably it depends on. I would let everybody decide if they. They are more of an audio or a visual type. Of course, the courses are newer and the coverage is usually deeper, so. So yeah, probably the book first.

C (34:05)

And how do you think the course and the book complement each other in terms of content.

D (34:10)

The book is the very quick one. So if someone has maybe just one day, then I recommend the book. And also it's easier to just go back to the book and take pieces from it by the course. It might need a little more time to let things think, try things I explained in the course.

C (34:32)

Well, that's a great feedback for anybody trying to decide between the book and the course, but I would say maybe probably both to just sharpen your skills. Well, thank you so much, Milan, for it's been a pleasure to know about this book and to dive deeper into it with you. Thank you so much for giving us time to discuss this and thank you to all our audience who tuned in for this episode and I look forward to seeing you guys in the next episode to discuss some more fascinating books on geography. Thanks, Milan.

D (35:05)

Yeah, thank you very much, Davika. It has been a great pleasure to be here.

C (35:09)

Pleasure to have you.