The past couple of months I had the pleasure to work for arguably the greatest Dutch global brand in retail. The goal of my work was to construct a geographical dataset for analysis and reporting. Luckily, geo-data and -information is a rabbit hole I’m familiar with. This is a short introduction to the world of Geo data to pique your curiosity.
Welcome to planet earth, we (not all) call it a globe. Fun fact: if you compress the earth to the size of a billiard ball, the earth is more smooth than the billiard ball.
In order to define a location on the planet many systems were used, and many still are. The most common system these days is probably WGS84 (also known as WGS 1984, EPSG:4326). And most commonly referred as “what Google maps uses”.
This WGS84 is used by GPS as well, that’s probably why Google chose it. To turn this into a flat plane they use web mercator. 
As you can see this creates an interesting effect, maybe the reason why so many people overestimate the size of Europe compared to Africa for example.
These pictures are taken from Cartography, Design, Research. 
This is just the tip of the ice globe, there are many more “uhm, what now?” moments when one dives into the world of geography.
This map is great for navigation but terrible for almost all other things. For now I just needed the coordinate system (for the Dutch speaking wordfeud players: coördinatenreferentiesysteem) to put customers and points of sales on the map. It’s great for that purpose.
Geocoding is the process of converting addresses into geographic coordinates, like Dinteloord to 51.635, 4.36944 where I am right now. The 51 latitude points more than half way up from the equator and the 4 longitude means I’m somewhat close to Greenwich .
Every now and then I run into a map with data near Mogadishu when people have the latitude and longitude mixed again.
Azure provides a batch service for Geocoding, something Google does not provide. We use this service to geocode CRM data.
Now we have customers and their sales as points on the WGS84 map, we want to do things like accumulating sales for the city of London, a great and typical European city in one of our most respected European countries that we like to have as a member in our union, our best island inhabiting friends, etc. etc.
But what is London? It is a concept. And concepts tend to change with the context they are in. As you can see in this visualization of the concept “B” or is it “13”?
This are polygons, sets of points connected to form a field. The Greater London polygon is actually two polygons, called a multi-polygon because it is donut shaped. Yum!
This is another “uhm, what now?” moment, our concepts of regions are not set in stone. London is fairly simple, Paris is chaos, Cyprus is even crazier sporting a UN Buffer Zone which does not belong to any country depending on who you ask.
When we have our data tied to absolute points on a coordinate system we like it is necessary to cluster those points, or aggregate them to a region like country. A country border is just a projection of a border concept and there are many public clusters you can find like postal codes, countries, streets, cities, etc. Or you sport your own maps based on service area’s, the weather, average age, income per household or cycle-time-to-nearest-store and this is where the data meets the business. What projection would you like to overlay on your data points?
Where art thou? H’re am I. 
You could say that I live in Dinteloord and municipality Steenbergen and the historical region of Brabantse Wal and the province of Noord-Brabant and The Netherlands and the Benelux and the European Union and Europe and planet Earth and the Sun system and the Milky Way galaxy and the Virgo cluster and the Virgo super cluster and the Universe and in some state of some string or just in a simulation.
Where are you?