I'd like to report the standard error of the clustering parameters (kappa, sigma) of an inhomogeneous Thomas point process model that I've fitted in spatstat. Yue and Loh (2015) reported doing this by a parametric bootstrap. I'm not very experienced in this concept, or applying it to point process models. How would I do this?
My first guess is to simulate my kppm a number of times and re-fit the resulting simulated points with the same covariates. Then, calculate the standard errors from the clustering parameters of each subsequent fitting. Is this correct? If so, how many simulations would be considered acceptable in this case? Thanks in advance for any pointers!
Basically your own description is completely correct.
My first guess is to simulate my kppm a number of times and re-fit the resulting simulated points with the same covariates. Then, calculate the standard errors from the clustering parameters of each subsequent fitting.
The only question left is how many simulations to do. Basically the answer is: "As many as you have time to do!". It is common to see people do 1000 simulations, so why don't you start there?