How to Anonymise Location Data
Highlights
- Geo masking where you actually move the points but in a random distance and direction but you maintain the spatial patterns still have to have enough observations for that but it gives you a little more leeway than simply up aggregating everything (View Highlight)
- ve maybe like an example of of how that could work I'll try to approach it backwards um so you want to have a situation where you look at a point on a map and it's been moved enough so that it could be any one of a group of points in your data set right so you cannot distinguish that point a is a the person you think it is among points A B C D and E a lot
of people like to have at least five different Neighbors that the point could be if you work backwards from that that will help you to determine the distance you're going to have to move points in order to make that happen we do something called a donut method so picture a donut right it's got a hole and then it's got an outer edge you have to move at least further than inside the hole so you have to move out to that boundary it's got to be in the donut so you don't have to go any further than the outer edge (View Highlight)
- o if you put it within the donut anywhere in that Circle right random Direction then you are in a safe Zone and when you move that point it's should if you have a relatively big enough data set doesn't have to be huge then you can maintain spatial patterns because the the five neighbors are still neighbors they've just all been moved so that you couldn't distinguish them (View Highlight)