An Introduction to Spatial Data & Geographic Information Systems

[MUSIC PLAYING]

NICK BEARMAN: I'm Nick Bearman, and I work at UCL as a teachingfellow in the Department of Geography,and I also run geospatial training solutions, whereI run training sessions helping people understand spatial data.[What is spatial geographic data,and how did you become interested in the field?]I'm really interested in how we canuse spatial data to solve some of the geographical problems

NICK BEARMAN [continued]: facing the world.Probably nearly every research projecthas got some element of spatial data, location data,within it--whether it be where people are, like in the census; wherepeople live; where people go to work; where they get shopping;how you interact with other people.It's a whole area of epidemiology,looking at disease and how diseasespreads through a population.

NICK BEARMAN [continued]: So it's a really wide field that cancover so many different areas.I was originally very interested in geography, and quitean interest in computers, as well.And this was kind of before GIS wasas mainstream as it was now.And it seemed like the real combination of the two areas.So the geography from kind of how

NICK BEARMAN [continued]: the world works, and computers allowing us to manipulatea whole range of data, and how we can use thattogether to try and understand the world better.[What is a geographic information system (GIS),and how can it be used?]GIS is geographic information systems,

NICK BEARMAN [continued]: or geographic information science,depending on which bit you want to use.And it's all about the technology behind mapsand behind spatial data.So the systems side is how we collectthe data, how we map the data, how we put iton the map, and the technology and the software behind that.So it can include GPS, include SATNAV, things like Google

NICK BEARMAN [continued]: Earth, that sort of thing.The science side-- geographic information science--is more the academic side.So how do we come up with a new methods?How do we store this data we've got as spatial data?How do we store the coordinates?What are coordinates?What's the projection system?How do we put these things on a map?And how do we go from our 3D globe

NICK BEARMAN [continued]: that we've got-- how do we put that on a flat, 2D map?GIS can be used to answer any question that involves where.So a lot of the areas that I work withwill be very interested in where people are.And this can be range of temporal scales.So whether it be like for a census--so every 10 years, we get a very detailed snapshot

NICK BEARMAN [continued]: of where people live and a lot of information about them,all the way through to much finer temporal information.So for example, we can use mobile phonesto track footfall in a shopping center.So how long do different people spend in certain shops?Which areas of the shops are more popular?Where are you most likely to see?And that, you can kind of get data every 30 seconds, almost.

NICK BEARMAN [continued]: [How have developments in technology improved GIS?]With census data, originally it was very much tabular-based.So computerized, but very little geography.With the developments in computer power,we can look at the spatial elementof the geography and the census data much more easily,

NICK BEARMAN [continued]: and gain much more information.But we're still probably looking at 100,000 data points,or something like that.So over the last kind of five years,we've come into much more of the big data area.So this can use data from a range of sources.So for example, we can do some very interesting analysiswith loyalty card data.

NICK BEARMAN [continued]: And for a kind of a big brand [INAUDIBLE] a loyalty card,you're looking at tens of millions of records.And we couldn't do that five years ago.We didn't have the computer power to do that.But now we can do some very interesting analysis with that.Look at spending patterns.And look at where people live versus where they go shopping,or whether they do click and collect.

NICK BEARMAN [continued]: Where do they pick up their purchases from?[What types of data can be collected and analyzed usingGIS?]You can get spatial data in many types.The census is one of the kind of key underpinning ones.Certainly in the last 20 or 30 years,the census has been a key kind of geographic data source.

NICK BEARMAN [continued]: But we also have a range of surveys.So the British Household Panel Survey is a very good source.It's very detailed, but the geography element of itis sometimes a little bit lacking, dependingexactly what you want to do.Because we don't have that many people in each geographic zone.We can also do some very interesting analysiswith Twitter data, looking at either geo-located tweets--

NICK BEARMAN [continued]: so these tweets people send with the GPS on their phone on.So it gives you a set of coordinatesfor where they sent that tweet from.Or looking at Twitter profiles to seewhere people put as their hometown,and to see how they're distributedand what sort of tweets people do.And Twitter is a great resource for this sort of thingbecause the data is all freely available.

NICK BEARMAN [continued]: It's all public, so we can do quite a lot of work with that.There's a whole range of different data sourcesthat we can use spatial data with.So this covers everything from thingslike the census and household surveys all the waythrough to LIDAR surveys and remote sensing.So the availability of data has increased dramatically

NICK BEARMAN [continued]: over the last five years.And even within the last two or three years, the availabilityof remotely sensed data--so satellite data-- is now quite amazing.And the detail that we can get from that is very high.So we can do lots of interesting environmental analysis lookingat, whether it be vegetation change, whether it be wildfires

NICK BEARMAN [continued]: and how they're spreading, even through to morekind of social science aspects.So there's a company launching whatthey call cube sats, which are very small satellites about 30centimeters by 10 by 10.So kind of very, very small.And they've got quite high resolution cameras.And they they're looking at can they

NICK BEARMAN [continued]: use them to monitor traffic in London,in terms of where cars are, and parking spaces, as well?So how does the traffic flow and whereare they most likely to get a parking space?So we can apply these sorts of dataacross the whole range of geography-- everythingfrom social science, human geography, all the waythrough to physical geography and environmental science.

NICK BEARMAN [continued]: [What computational skills and methods do you need whenworking with large scale data?]The biggest skill anyone can get if they'relooking to do this sort of thing is coding of some description.Because there are a lot of GIS tools out there.But to use some of these new data sources,we need to be able to create these tools, as well.

NICK BEARMAN [continued]: So the skills of coding are key to this.Probably the two biggest languages to knowwould be R and Python.And they both can do some really good data sciencewith geographic data.So I would recommend to anybody interested in that area,have a look at learning one or both of those.And even if you don't think of yourself

NICK BEARMAN [continued]: as a programmer or a coder, you can get into it and learn it.They're not very hard to learn.The kind of learning curve is a little bit steep,but it's well worth it.Because it allows you to do so much more with the datasets we have available.There are lots of sorts of analysisyou can do with spatial data.One of the kind of key building blocks

NICK BEARMAN [continued]: is choropleth maps, particularly lookingat rates of different variables.So disease is quite common one.There's a whole range you can do with that.And so understanding how those work,and how you can create a choropleth mapis a great way of communicating data.And it also explains some of the kind of theory

NICK BEARMAN [continued]: behind GIS, and why we need to collect data in certain ways.Heat maps do come up quite frequently.And they're a very useful way of gettinga quick overview of the data.So they can make very pretty maps quite quickly.But there's a limit to what you cando from a scientific point of viewwith those, in terms of interpreting them.

NICK BEARMAN [continued]: [What are the key things to keep in mind when visualizing data?]When you're visualizing geographic data,it's very important to make sure youknow something about the data set you're visualizing.So where does it come from?And what can you do with it?Because we collect data sets differently,

NICK BEARMAN [continued]: depending on what we're looking for.And so you can only use certain data sets for certain things.So rather than just finding some data on the internetand chucking it on a map, you need to actuallyhave a look at the data.Think about what is it showing?What is it collecting?So being critical is very key to that.Another thing to be aware of is, how old the data are.

NICK BEARMAN [continued]: Again, it depends on what you're trying to show, what messageyou're trying to communicate.But being aware of how old the data might be, how up to dateit is, is quite important.Another thing to bear in mind about data visualization,particularly with choropleth maps,is to make sure you're visualizing rates,or say, a rate of disease per 1,000 population,

NICK BEARMAN [continued]: or whatever it might be, rather than just the count of people.Because if you're trying to show how the rate varies across,for example, the UK, if you just see the number of cases,then we just get a map of population.It doesn't tell you anything, you know.It tells you most of the people are in big cities,say London or Birmingham.Which, we know that already.So it's really important to make sure to use

NICK BEARMAN [continued]: the rate, or the per capita, for the data.[How do you manage large scale geographic data?]When working with geographic data,it's very easy to end up with lots of data.So it's good to have a system in place for managing that.There is a list as long as your arm of different formats

NICK BEARMAN [continued]: of spatial data.And in some ways, the format you useisn't too vital, as long as you know what the data are.And probably the best thing I can recommend to anybodyis make sure you have a good filing system in your MyDocuments folder.So have different folders for different things.And make sure you know what you're saving where.

NICK BEARMAN [continued]: Spatial data is normally stored as files or databases,so files are probably the easiest way to start off with.But as soon as you get into any bigger projects,you'll be ending up doing stuff with databases.So if you want to go into the area,PostgreSQL is probably one of the most common ones.So that will come up a lot.So it's worth having a look at that and seeing how that works.

NICK BEARMAN [continued]: It's not difficult to set up.And it's just a different way of thinking about dataand how it's stored.So again, spend just a bit of time thinkingabout how it's stored, how it's organized.And you can get a long way with that.And databases are going to become much more prevalent whenwe're looking at accessing data through the internetand real time data.Because 95% of data on the internet

NICK BEARMAN [continued]: is stored in databases of some form.So accessing that database and using that data is a key skill.[Tell us about your project "PopChange."]So one of the projects I've been working on recentlyis a project called PopChange, wherewe are looking at census data.But we want to do small-area comparisons over time.

NICK BEARMAN [continued]: In the UK, the smallest area of census data that we getis called an output area.And there are about 100 people or so, give or take.But over the last kind of five censuses-- so 2011, 2001,all the way back to 1971--these changed slightly.Sometimes quite dramatically.Sometimes only slightly.But it makes comparing these small areas over time

NICK BEARMAN [continued]: quite difficult.So in the PopChange Project, we reallocated populationsfrom our output area geographies to a one-kilometer gridacross the whole country.And that's a consistent grid over time.So that allows us to do this small area comparisonto see how a smaller community, or a small area of the city,

NICK BEARMAN [continued]: has changed over these last 50 years.The data management for that was quite intensebecause of the amount of data that we've had.And we used a mixture of databases and files for that.And one of the key skills for creating that was using R.

NICK BEARMAN [continued]: So we used the R programming language to process the dataand manage the data.So that's a really important aspect.[Did you learn anything unexpected while workingon this project?]Whenever you're dealing with the area of computer science,even from a GIS point of view, where

NICK BEARMAN [continued]: we're using computer science in combination with others,the rate of change is quite amazing, really.And the different technologies that come across,they change very rapidly.So we were working with a software developer,and he used a language called Closureto create the web interface for the PopChange Project.

NICK BEARMAN [continued]: I'd never come across that before.It's one of many languages out there.And it was quite well-suited for what he wanted to do.But he'd actually also only used that language a few times.So as a part of the project, it was learning how that works.But he's proficient in a whole range of other languages,so it's very easy for him to adapt.And that's the common theme throughout, certainly,

NICK BEARMAN [continued]: geographic data science and GIS, as well.You'll always get new technologies and languagescoming along.So you always keep learning.So I've been using R for quite a few years now.But I've started to do more with Python now.And no doubt, in another two or three years,there'll be another language that comes along,

NICK BEARMAN [continued]: or another way of doing things.So part of the key element of thisis to make sure you keep your skills up to date,and sort of see what the new things arethat are coming about.And you will end up spending time having to learn them.But that goes for everyone.Nobody ever stops learning in this area.[What tools would you recommend using for this type

NICK BEARMAN [continued]: of research?]There's a whole range of different toolsthat we can use to create maps.One of the things you'll get used tois how different tools work.So I can really recommend trying different tools,whether that will be just having a website

NICK BEARMAN [continued]: and see how they work.If there's a training opportunity that comes alongto your university or department or wherever you're based,go along to that and see how it works.And spend some time playing about with the software.One particular tool I found really useful,particularly for choropleth maps is somethingcalled ColorBrewer.And that's a website that helps you choose colors that

NICK BEARMAN [continued]: work well for choropleth maps.And other very common tools are becomingvery popular, particularly open source ones.So we've mentioned R already.QGIS is a very good open source GISprogram that you can run on your desktop.So I'd really recommend checking that out.And there are a whole range of web-based tools

NICK BEARMAN [continued]: for doing web maps.So that Mapbox and Leaflet are very, very common ones--very popular.And CARTO is one that's come on the scene relatively recently,and is worth checking out.But there will always be more tools coming along.So there's no such thing as a definite list of tools.There will always be new things to add.[What advice would you give people looking to do research

NICK BEARMAN [continued]: using GIS?]If you're quite new to using spatial data,the first thing I'd recommend is,have a look at Google Earth, if you haven't already.It's an amazing piece of software.And when it was launched, it revolutionized how everyoneperceived spatial data.Because before that, it was quite

NICK BEARMAN [continued]: difficult to get hold of spatial data.You needed very specialist software.When that came out, it allowed anybody to look at the globeas it is, and zoom in on different locationsand see what their local area looked like,see what their house looked like.And that is a very basic GIS.It doesn't have all the kind of editing capabilities.But it still manages spatial data.

NICK BEARMAN [continued]: So that's a great resource to start with.And it's also worth thinking about,what data do you have in your field?Because with the research you've done already,you will be using data.I guarantee that you'll do something with data.And 90% of all data has got some spatial component.So it might be nice and easy, like an address, or a root,

NICK BEARMAN [continued]: or a location of data that you've collected.And that we can put on a map very easily.It might be something a bit more not so directly spatial.So if you're doing textual analysis,you might be looking at location names, different locations thatare mentioned in a book, or something like that, perhaps.Or you might have to think a little bit more laterally.

NICK BEARMAN [continued]: And so if you're dealing with customers,where do they use your service, if that's the area?And where are they based?How do they travel to get to your service?So there's a whole range of different data setsthat have a spatial element.And you know the data set you're using very well.

NICK BEARMAN [continued]: So there's probably some kind of geographic element of that.And you can bring those skills to the tablewhen you work with some working in GIS,or when you create your own map.So there's a whole range of different skillsyou can bring to that.[MUSIC PLAYING]