Symbol location accuracy using addresses/place names
Often when making a symbol map, we want to use specific geographic locations such as counties, cities, or even neighborhoods. In order to create a symbol map on Datawrapper, you need to upload your data with either
- Address/Place Names or
- Latitude/Longitude.
Since most of us don't always have specific latitude and longitude data, we use place names to create our map. In this tutorial, we'll review how to ensure your symbols appear in the correct location on the map using address/place names.
Create a map using addresses and place names
For this example, we'll be creating a map of New York state's counties and how many farmers' markets each has. The first step in the process is to select the base map - in this case, New York State counties. Easy enough.
The next step is to add our data. This is where it can get tricky. When adding data using place names, try to be as specific as possible so that our geocoder is better able to locate places (though exact addresses are not necessary).
What is geocoding?
Geocoding is the process of automatically mapping a description of a location—such as a pair of coordinates, an address, or a name of a place—to its geographical coordinates. Sometimes it may seem like you get unexpected or unpredictable results from the geocoding process. Geocoders can't always be 100% reliable as there are often multiple places with the same name, and it doesn't know which one you're aiming for.
Think of it like when you put an address into Google Maps - there are always multiple results, and the one you had in mind isn't necessarily the first one. A geocoder works the same, except you have no chance but to look through the list of search results to determine which one is correct - the geocoder always takes the first result. Again, it is helpful to be as specific as possible in the address definition.
Here are a few quick-reference examples for different types of locations:
- Neighborhood: Williamsburg, Brooklyn, New York
- City: New York City, New York
- Zip Code: 11212, Kings County, New York
- County: Kings County, New York
- State/Land: New York State, USA
Examples
Let's take a look at three outcomes using our New York counties example in order from least accurate to most accurate:
Using Only County Names
Result #1:
In this example, the county name is ambiguous. Kings, for instance, could be a county, city, or neighbourhood anywhere in the US. It is visible in the map that many counties are missing symbols, and there are symbols located in the white space. This is also indicated in the Check tab as unmatched symbols (highlighted in yellow). The number of errors makes this map completely unusable.
Result #2:
In the second example, the county name + the state name is clearer, but again, Kings, New York could be either a town or a county (which happens to be a town in upstate New York). If you look closely, several counties are missing symbols and some counties have more than one symbol. This is again indicated under Check tab. This map is really still not ready for use.
Using County + the word 'County' AND State Name
Result #3:
The third map is perfect. In the last example, county name + "county" + state name ( Kings County, New York) ensures that the geocoder knows we are referencing a county and is able to place all the symbols correctly with 100% accuracy. Feel free to zoom in on the maps and see if you can locate the errors.
In summary, when creating a symbol map, be sure to include specific location names when importing your data using addresses/place names. The more specific you are, the more likely the geocoding process will deliver accurate symbol placement.
Certain maps in which we have moved islands closer to the mainland country (e.g. Hawaii, USA or the Canary Islands, Spain) will cause additional issues with the geocoding process.
For example, locating the Canary Islands by place names will not work because we shifted the Canary Islands closer to mainland Spain for aesthetic purposes. Our latitude and longitude for the Canary Islands are therefore not exactly "true." Unfortunately, place names will not work because they rely on latitude and longitude. At the moment, we have no possibility to offer a map with the Canary Islands close to the mainland with the correct coordinates.
A solution to this problem is to use latitude and longitude with adjusted values instead of addresses or place names. For example, for all the data points located on the Canary Islands, add the values 7 to your latitude and 3 to longitude, and for the Galapagos Islands, add the value 0 to your latitude and to your longitude. As an example, for Las Palmas, we need to add 7 degrees to the latitude and 3 to the longitude so that Latitude 35.15 + 7 = 42.15, Longitude -12.43 + 3 = -9.43. The new latitude/longitude for Las Palmas is then 42.15,-9.43. This will cause the symbol to be in the correct location.