Anaphory
commented
2 years ago Minimal example:
> minimal_network = sfnetwork(tibble(x = c(1,2,3), y=c(2,1,3)) %>% st_as_sf(coords=c("x", "y")), edges=tibble(from=c(1,2), to=c(2,3), attribute=c("a", "b")))
Checking if spatial network structure is valid...
Spatial network structure is valid
> minimal_network
# A sfnetwork with 3 nodes and 2 edges
#
# CRS: NA
#
# A rooted tree with spatially implicit edges
#
# Node Data: 3 × 1 (active)
# Geometry type: POINT
# Dimension: XY
# Bounding box: xmin: 1 ymin: 1 xmax: 3 ymax: 3
geometry
<POINT>
1 (1 2)
2 (2 1)
3 (3 3)
#
# Edge Data: 2 × 3
from to attribute
<int> <int> <chr>
1 1 2 a
2 2 3 b
> convert(minimal_network, to_spatial_smooth)
# A sfnetwork with 2 nodes and 1 edges
#
# CRS: NA
#
# A rooted tree with spatially implicit edges
#
# Node Data: 2 × 2 (active)
# Geometry type: POINT
# Dimension: XY
# Bounding box: xmin: 1 ymin: 2 xmax: 3 ymax: 3
geometry .tidygraph_node_index
<POINT> <int>
1 (1 2) 1
2 (3 3) 3
#
# Edge Data: 1 × 4
from to attribute .tidygraph_edge_index
<int> <int> <chr> <named list>
1 1 2 NA <dbl [2]>
agila5
luukvdmeer
Is your feature request related to a problem? Please describe.
I start with a network and then try to turn it into a simple, smooth network. This needs to be done iteratively: Imagine a network section like this.
Only smoothing it woud remove 2,4,5; only simplifying it would remove the edges around 6; but iteratively smoothing then simplifying it will remove all down to a
1-7connection. For shortest path routing, I don't need all those edges, just the shortest connection.However, while I can tell
to_spatial_simplehow to merge attributes (summarise_attributes = "first"),to_spatial_smoothhas no such optional argument. Instead, I get NAs in my edge attributes.Describe the solution you'd like
with the same semantics for
summarise_attributesas into_spatial_simple.Describe alternatives you've considered
I could just not smooth my graph. I start with 24951 nodes and 29292 edges and end up with 19920 nodes and 23980 edges, so it's quite a chunk.
PS: Very minor thing: While trying to understand how this could be implemented, and why it's not happening at the moment, I noticed that this check https://github.com/luukvdmeer/sfnetworks/blob/10b009e9d112d50c4f95c326b92864c1eb4f6ba6/R/morphers.R#L589 happens in every row for lapply. I think you could assume that the graph doesn't change directedness inside the lapply, and wrap the two different find_edges functions into the two different branches of that
if?