Adding a new type for storing point patterns #46
Conversation
|
Thank you @JakeGrainger for looking into it. Can you please clarify the use cases you have in mind? Perhaps we already have everything we need in the GeoTables.jl package, which will be clear in the book that is coming up soon: https://discourse.julialang.org/t/new-book-geospatial-data-science-with-julia/103364/1 Basically, we attach any table to a PointSet to produce a marked process. My understanding is that we don't need to introduce a new type for that. Can we postpone this discussion to after the book is out? You will probably have a better idea of how we envision these concepts together. |
|
Ah sorry I hadn't seen that, yeah for the marks that solves it so this isn't necessary, and GeoTables .jl is much more general as an interface right so that seems nicer. |
|
We can certainly brainstorm all these ideas after the book is out (next
week or 2 weeks from now).
It would be nice to hangout more in our GeoStats.jl community channel on
zulip to build these concepts together.
Em ter., 26 de set. de 2023 11:06, Jake P. Grainger <
***@***.***> escreveu:
… Ah sorry I hadn't seen that, yeah for the marks that solves it so this
isn't necessary, and GeoTables .jl is much more general as an interface
right so that seems nicer.
So for example lets say we want code for thinning a marked point pattern,
do you envisage that being added here and adding GeoTables as a dependency,
or the other direction?
Also, I guess the only remaining difference is that the region on which
the points are observed is not currently packaged with the points, but I
guess you can pass that around separately.
However it does seem like there can be benefits to keeping them together,
which is I think what spatstat does for example.
I guess this is more of an issue if this package will have analysis
methods as well as just simulation.
—
Reply to this email directly, view it on GitHub
<#46 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAZQW3P5BWGI44YMYCBUOWDX4LONDANCNFSM6AAAAAA5HXF7W4>
.
You are receiving this because you commented.Message ID:
***@***.***>
|
|
Yeah definitely, look forward to seeing the book. But yes I think given the GeoTables stuff this pull request is redundant |
I was just thinking that whilst the PointSet interface is nice, many algorithms also require the observational domain, and potentially marks for the process. So I have added a PointPattern type which can store these and modified the existing simulation methods to return these. I have tried to be minimal with it but see what you think?