Skip to content

Spatial Query (Selection) with Apply

4 messages · Robin Lovelace, Ariel Fuentesdi

Ariel Fuentesdi
# 
Hi,

I want to do multiple selections of a point shapefile based on polygons on
other layers, I can do this in a for loop, but I desire to do this in a
function of the apply family.

I named the point shapefile "nodes" and the polygons shapefile "zones".

This is what I did:

tmp <- list()
for (i in 1:nrow(zones at data)) {
    tmp[[i]] <- nodes[subset(zones, ESTUDIO == i),]
    tmp
  }

But I have no clue how to change it to the apply family, can you provide
an example of this?

Thanks in advance.

Regards,
Ariel Fuentes
Robin Lovelace
# 
Hi Ariel,

It helps when asking for help with code to produce a reproducible example.
To understand your input data I've created nodes and zones based on the
spData data nz_height and nz_elev. Based on the assumption you want to find
all the nodes in each zone I think the direct answer to your question is
something like the following:

tmp2 = lapply(s, function(x) {
  nodes[zones[x, ], ]
})

The longer answer is that aggregate(), st_join() + aggregate()/summarize()
may provide quicker solutions, depending on what you want to do with the
points after grouping them by which zone they fall in.
Note: I've used sf objects based on the explanation here https://geocompr.
robinlovelace.net/spatial-operations.html#spatial-vec which may not work
with sp data but the aggregate code should work roughly the same:

# install.packages("sf")
# install.packages("spData")
library(spData)
library(sf)
#> Linking to GEOS 3.6.2, GDAL 2.2.3, proj.4 4.9.3
zones = nz
nodes = nz_height
tmp = list()
s = 1:nrow(nz)

for(i in s) {
  tmp[[i]] = nodes[nz[i, ], ]
}

# understand what's going on with plots (not shown)
# plot(st_geometry(nz))
# plot(st_geometry(tmp[[i]]), add = TRUE, col = "red")
# plot(nz[i, ], col = "green", add = TRUE)

tmp2 = lapply(s, function(x) {
  nz_height[nz[x, ], ]
})

identical(tmp, tmp2)
#> [1] TRUE

Created on 2018-08-27 by the [reprex package](http://reprex.tidyverse.org)
(v0.2.0).

I've also pasted the reprex into the geocompr github tracker so the plots
can be seen and the code formatted: https://github.com/
Robinlovelace/geocompr/issues/294

Hope this helps,

Robin


On Mon, Aug 27, 2018 at 9:53 PM, Ariel Fuentesdi <ariel.fuentesdi at usach.cl>
wrote:

  
  


  


      

    

    

      
      

        


  

        

      Ariel Fuentesdi
    

    

      
      #
    

  


  

      
Thanks, Robin

It worked perfectly, although I transformed the data from sf to
SpatialDataFrame objects with:

zones <- as(zones, "Spatial")
nodes <- as(nodes, "Spatial")

The next step I will follow is to create Distance Matrix like this:

coord <- tmp2[[1]]@coords
dist1 <- apply(X = coord, MARGIN = 1, FUN = function(x) spDistsN1(coord, x,
longlat = T))

obviously, the task is a distance matrix for every selection. ?What would
you recommend, the aggregate function as you talked before or something
like a nested apply?

Regards,
Ariel



2018-08-27 19:32 GMT-03:00 Robin Lovelace <rob00x at gmail.com>:

            

      
      
      
    

        

      
      
    

            

      
      
      
    

        

  
  


  


      

    

    

      
      

        


  

        

      Robin Lovelace
    

    

      
      #
    

  


  

      
Hi Ariel,

Glad it helped.

Yes you could place your apply command into a lapply to get a list of
matrices.

Something like this should work:

dist_list <- lapply(tmp, function(coord) {

   apply(X = coord, MARGIN = 1, FUN = function(x) spDistsN1(coord, x,
longlat = T)

}

Robin



On Tue, Aug 28, 2018 at 9:13 PM, Ariel Fuentesdi <ariel.fuentesdi at usach.cl>
wrote: