Meet the team: Chris van Diemen on algorithms and visualizing results
Green City Watch’s Digital Communications Intern, Anna Roberts, recently sat down (virtually) with Lead Data-Scientist Chris van Diemen to discuss algorithms and how to visualize their results. Curious about the people behind Green City Watch? Keep on reading! Interviews have been edited and condensed for clarity.
Anna: To start off, where are you from?
C: I was born in Larnaca which is in Cyprus. But I was raised in the North of Holland, in a small town called Hoogkarspel next to Lutjebroek.
A: What are your interests, generally?
C: Generally, I’m quite interested in technology and nature and the interplay between those two.
A: How do you think you first became interested in this?
C: First? That’s a good question. Probably a long time ago… My father is a plant breeder so he creates new species of plants.
A: Very cool.
C: He was actually one of the first to do plant breeding according to the organic rules, without using artificial fertilizer, insecticides, and stuff like that. He was always working with plants so I think I got the nature part from him. And the technology, I don’t know. I remember a video of me where my mother is making a recording with one of these old-school, hand-held cameras. Anything with buttons always had an immense impact on me. I really was drawn to it. I don’t know why that is but it kind of stuck with me. So anything with buttons, I needed to pull it apart.
A: That’s very sweet.
C: Yeah, it’s a funny video. I’m asking my mom, “Can I hold the camera?”, but she doesn’t trust me with it. And rightfully so, I think.
A: So I guess at Green City Watch, you’re trusted with a bit more things now.
C: Yeah, I like to think so.
A: Would you state what your role is at Green City Watch?
C: I am the lead data scientist.
A: And what does that mean?
C: So Jim, our CTO, comes in to make sure we have the right data, that there is information in the data that we need. I take the data and try to extract the information that we need in a scalable way. So a practical example that we are working on now: a tree reflects light differently than other things. There are existing tools where we can select images that reflect a certain spectrum of light and say, “That is most likely a tree”. And then what I do is I try to automate that process. I give the computer a lot of these trees, say “these are all trees”, and the computer can learn that it is a tree. Now we don’t have to detect every tree by hand and we can let the computer do the work. So if we want to know where each tree is, just put in a new image and let the computer decide where the trees are.
A: And that’s machine learning, right?
C: Yeah, there are many names. These days you can say machine learning, big data, AI, algorithms. But to me, many of these terms are describing similar things.
A: We will be discussing algorithms as the main subject for this interview. Would you mind explaining how you became involved with algorithms or how you first began working with them?
C: I studied Future Planet Studies at the University of Amsterdam for my bachelor’s program. They divide the program into beta and gamma. So the gamma part is more about the politics of environmental sciences and the beta part is about understanding the climate models. Once you get into the climate models, there is a lot of information needed to make these models work. You need to know the land cover of the world and the flow of chemicals in the atmosphere. A lot of methods have been developed to measure these things or proxies of these things using remote sensing. When you want to measure proxies or when you want to measure land cover, for instance, you run into the scalability issue. You want to be able to measure the land cover for a huge area, the whole planet ideally. That’s when I first started to encounter the algorithms, where you want to make a set of rules that’s mathematically described. That’s what an algorithm is to me. That makes it easier to do analysis on a large scale.
A: So I’m sure that climate models are one environmental use of algorithms. Do any other “green” uses of algorithms come to your mind?
C: What we are working on today is to measure the trees. Where are the trees and how are they doing? We want to improve life on Earth, and we want to especially focus on life in cities. Because more than 60% of all human beings will be living in cities soon. Increasing the tree cover is actually an improvement on so many fronts. It is a very useful thing to apply these scalable algorithms to determine how the trees are doing. As Nadinè says, cities are undergoing a digital revolution. If we don’t ‘take nature online’, its management might lag behind. That’s also getting into the visualization; we can use these algorithms to map trees and how they are doing. By putting that online, people will use it, and it will have an impact on how these things are valued. I think people kind of want this information. They want to know how the trees are doing. As for now, it’s very hard work to do because you need a lot of time and knowledge. We will need that knowledge in the future, but we can speed up the process of some basic tasks in determining where trees are and giving an early warning system for if the trees are in trouble. That is what we can use these algorithms for, which is very useful at the moment.
A: You mentioned Green City Watch is using algorithms to identify trees and assess their health, and then creating a map to visualize these results. Is this the main way of visualizing an algorithm’s results or are there other key visualizations?
C: I think that it’s really, kind of an art. This is what major companies are spending billions and billions of euros on and then it is not about trees but about products. The way you visualize and the way you present things to people, that’s one of the largest industries, I think. It’s really a challenge to do it in the right way. What we do now is a map because we really love maps.
A: Me too.
C: I think they are visually very attractive. But I think we have to experiment in different ways in the near future. What we really try to do is work together with the clients and see how they use these tools. And by doing that, we can focus on stuff that we like as technologists and map-enthusiasts to what people actually need in the business to make their decisions. But I think in this field, many of these are displayed in dashboards.
A: What does that mean?
C: A dashboard is a way to interact with data. So if you have some data that has values over time or values in space, you can have a dashboard, which means you have some graphs on a screen that gives you insights into your information. In our case, on this dashboard, we have a map where you can see, spatially, where your information is in the world. We also have some graphs about how it might develop over time. Is there an increase over time, or a decrease? That’s what a dashboard can convey to you so you can have different information in one glance.
A: Very interesting.
C: This is also changing these days because many people realize that if they have a dashboard, they might need more information. Actually, when you get more information, you get more questions. So what that means, is that there is a bit of a move to use notebooks. This is a page where you have some sort of code that can be Python, which is a programming language, but it can also be much easier with a drag-and-drop system. You can have the variables that you’re interested in and combine them in a way that makes sense to you. We are basically looking for ways to bridge that gap between people who are very tech-savvy and want to write their code and people who maybe don’t want to write code but still want to have modifications to the analysis they’re looking at.
A: Wow. Just to go back real quick to the map enthusiasm, why would you say this is a good form of visualization?
C: I think you can do many things with maps. Just how we evolved, we have quite a good spatial sense. That’s why I think it really makes sense to use maps. Because people can look at a map and think, “Ok, I’m here. And these things are close to me or are somewhere in the world where I can go.” With maps and colors and nowadays, the 3-D visualizations, there is much more that you can communicate with a map than just location. What is also interesting, is the biggest apps on mobile phones, are Google Maps or the Waze for navigation. I think in that way, maps have really become more prolific in people’s lives in the past couple of years. One thing I notice about Google Maps is that the green areas are actually quite explicit features.
A: You mean that they’re very distinct?
C: Yeah I think if you look at the Google Maps, and not the lines feature but just the area features, there is green, there is grey, and there is…
A: Blue.
C: Shopping areas [laughs]. Those are the main features. It is interesting to look at a city, and as you zoom in, the green areas stand out more. I think that’s very nicely done. And it’s not necessarily because Google likes green areas more than shopping areas, but because people use these green areas, so they are pretty clear on maps. It’s just an observation.
A: It’s a nice one. It’s also very similar to the project you all worked on with Amsterdam, the #UrbanNatureAmsterdam map?
C: Oh yeah, that’s a beautiful example. But that takes it a step further and really emphasizes these green areas.
A: Do you have any final things you’d like to add?
C: That’s it from me.
A: Thank you for doing this interview.
This interview is part of the “Meet the Team” project, aimed at showcasing the motivated team behind Green City Watch, while also spreading information about important topics. Each team member is asked to think of a subject that they find most interesting and is subsequently interviewed about their perspective and opinions. Want more? Read Jim’s interview here.
