2
$\begingroup$

I have a spatial point pattern which spans across multiple districts in the country.

Now I want to examine its pattern using Ripley's L function.

I know that Ripley's L function is used for completely mapped data analysis.

Therefore, I want to know what is your recommendation in choosing the study area.

Should I just simply consider the polygon which represents all the districts as my study area or is it better to choose minimum bounding rectangle or even sample the data?

$\endgroup$
1
$\begingroup$

Normally it should be the combined area of all the districts where you have fully sampled the pattern you are trying to analyze. In many cases this will be a single polygon, but it can easily be a polygonal area consisting of the districts you have data from and holes in between from areas that have not been surveyed.

$\endgroup$
  • $\begingroup$ in my case, the point pattern represents the damage caused by earthquake. As this work (assigning the damage) was done by volunteers, I want to examine its pattern. But I am not sure how to figure out in which district the patterns is fully sampled and in which not ... $\endgroup$ – MichalB May 20 '16 at 16:43
  • $\begingroup$ An alternative is to simply use the entire dataset you have, but then you are modelling the point process of "reported damages" and not the real point process of damages. You might still find some interesting information though. $\endgroup$ – Ege Rubak May 23 '16 at 21:07
  • $\begingroup$ May you please clarify more what you mean by "reported damages" vs. "real damage"? Thanks. $\endgroup$ – MichalB May 24 '16 at 5:12
  • $\begingroup$ Well the data you have is literally the locations of damages that somebody took the time to report, so this is what you have for sure. In an ideal world this is equal to all the damages (so a realization of the process of damages), but maybe in some areas the surveyor forgot to report half the damages, so you only have data on the process after a "thinning". In point process theory a thinning is when you delete some of the points of the process. If you have a model for how the thinning is done you might still be able to say something about the original process. $\endgroup$ – Ege Rubak May 25 '16 at 16:16

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.