Thread by @WeAreRLadies

So - #rspatial work in insurance! Other than natural perils (which I will talk about at some point) it’s also used to identify the features of individual dwellings that might impact their risk of things like burglary or malicious damage.

For example, I found every house in Australia that’s in a culdesac (Less people going through might reduce theft, but access for fire engines is harder and increases fire risk). Here’s a quick screenshot of visual checks I did using the package leaflet to make sure it was working

I really like doing this kind of stuff in R, because you have to think about things in a very different way than with most other types of analyses.
Coming back to my example, how would you even begin to find every dwelling that’s in a culdesac?

We have every road in Australia as a Spatial Line (a type of spatial data that you can create & deal with using the sp package). Every time lines intersect, a new line starts. So at an intersection, you have 4 line segments coming together in the middle

We can extract the end point of each line & add a smaller buffer to those end points (done using the gBuffer function from rgeos). Now we can test whether each buffered end point intersects with any other end points. If it does - the road continues! If not, it’s a cul-de-sac!