For my thesis I'm trying to create a method somewhat similar to Singleton et al. (2011) to calculate the catchment areas of schools. This method uses a combination of 2D-kernel density estimation and percent volume contours. In this method the bandwidth is increased by 100 meters everytime the percent volume contour is not a single unified polygon. I have been trying to recreate this with Python. Later this will be implemented in Arcmap 10.6.1. The problem is that the Python functions do not express their bandwidth in meters but instead by a bandwidth factor. I have been using scipy.stats.gaussian_kde().
My datapoints (the addresses of students) have x- and y- coordinates that are expressed in meters and the coordinate systems of my dataframe in Python and Arcmap has meters as units too.
So the question is: how do I calculate what the bandwidth is in meters?
Note: I am aware that Singleton et al. (2011) uses a different type of kernel function, but the Gaussian function should also be applicable in this method.