sum_network_routes results differ between sf and sp data

[Robinlovelace]

Identified here: https://github.com/Robinlovelace/geocomp-tranport-workshop

If

ways_sln = SpatialLinesNetwork(ways_sp)

is replaced by

ways_sln = SpatialLinesNetwork(ways)

It produces an incorrect result.

[mpadge]

If ways_sp is constructed as as (ways, "Spatial"), then I get the same values

ways_sp <- as (ways, "Spatial")
ways_sln2 <- SpatialLinesNetwork (ways_sp)
identical (slot (ways_sln, "g"), slot (ways_sln2, "g")) # FALSE, but only coz of igraph hash
identical (slot (ways_sln, "nb"), slot (ways_sln2, "nb")) # TRUE
identical (slot (ways_sln, "weightfield"), slot (ways_sln2, "weightfield")) # TRUE

looks good to me

[Robinlovelace]

Can you reproduce this plot? The results suggest it's not a big issue, something linked to the creation of the sf object I think which should be relatively straightforward to fix, hopefully. Even better: it's not a bug and only affect me (I'm using the dev version of splanr but cannot think of changes that could have introduced this regression):

library(stplanr)
from = "Jena Paradies"
to = "JenTower"
(from_coords = geo_code(from))
#> [1] 11.58766 50.92493
(to_coords = geo_code(to))
#> [1] 11.58465 50.92883
u = "https://github.com/Robinlovelace/geocomp-tranport-workshop/raw/master/ways.geojson"
if (!file.exists("ways.geojson")) {
  download.file(url = u, destfile = "ways.geojson")
}
ways = sf::read_sf("ways.geojson")
ways_sp = as(ways, "Spatial")

ways_sln = SpatialLinesNetwork(ways_sp)
from_sln = find_network_nodes(ways_sln, from_coords[1], from_coords[2])
to_sln = find_network_nodes(ways_sln, to_coords[1], to_coords[2])
r_local = sum_network_routes(ways_sln, from_sln, to_sln, "length", combinations = F)

ways_sln = SpatialLinesNetwork(ways)
#> Linking to GEOS 3.5.1, GDAL 2.2.2, proj.4 4.9.2
from_sln = find_network_nodes(ways_sln, from_coords[1], from_coords[2])
to_sln = find_network_nodes(ways_sln, to_coords[1], to_coords[2])
r_local_sf = sum_network_routes(ways_sln, from_sln, to_sln, "length", combinations = F)

plot(r_local, lwd = 5)
plot(r_local_sf$geometry, add = TRUE, col = "red")

Session info

devtools::session_info()
#> Session info -------------------------------------------------------------
#>  setting  value                       
#>  version  R version 3.4.4 (2018-03-15)
#>  system   x86_64, linux-gnu           
#>  ui       X11                         
#>  language en_GB:en                    
#>  collate  en_GB.UTF-8                 
#>  tz       Europe/London               
#>  date     2018-04-11
#> Packages -----------------------------------------------------------------
#>  package     * version    date       source                           
#>  assertthat    0.2.0      2017-04-11 cran (@0.2.0)                    
#>  backports     1.1.2      2017-12-13 CRAN (R 3.4.3)                   
#>  base        * 3.4.4      2018-03-16 local                            
#>  bindr         0.1.1      2018-03-13 CRAN (R 3.4.4)                   
#>  bindrcpp    * 0.2.2      2018-03-29 CRAN (R 3.4.4)                   
#>  class         7.3-14     2015-08-30 CRAN (R 3.4.0)                   
#>  classInt      0.1-24     2017-04-16 cran (@0.1-24)                   
#>  compiler      3.4.4      2018-03-16 local                            
#>  curl          3.2        2018-03-28 CRAN (R 3.4.4)                   
#>  datasets    * 3.4.4      2018-03-16 local                            
#>  DBI           0.8        2018-03-02 cran (@0.8)                      
#>  devtools      1.13.5     2018-02-18 CRAN (R 3.4.4)                   
#>  digest        0.6.15     2018-01-28 CRAN (R 3.4.3)                   
#>  dplyr         0.7.4      2017-09-28 CRAN (R 3.4.3)                   
#>  e1071         1.6-8      2017-02-02 cran (@1.6-8)                    
#>  evaluate      0.10.1     2017-06-24 CRAN (R 3.4.1)                   
#>  foreign       0.8-69     2017-06-21 CRAN (R 3.4.0)                   
#>  formatR       1.5        2017-04-25 CRAN (R 3.4.1)                   
#>  geosphere     1.5-7      2017-11-05 CRAN (R 3.4.2)                   
#>  glue          1.2.0      2017-10-29 CRAN (R 3.4.2)                   
#>  graphics    * 3.4.4      2018-03-16 local                            
#>  grDevices   * 3.4.4      2018-03-16 local                            
#>  grid          3.4.4      2018-03-16 local                            
#>  htmltools     0.3.6      2017-04-28 cran (@0.3.6)                    
#>  httr          1.3.1      2017-08-20 cran (@1.3.1)                    
#>  igraph        1.2.1      2018-03-10 cran (@1.2.1)                    
#>  jsonlite      1.5        2017-06-01 cran (@1.5)                      
#>  knitr         1.20       2018-02-20 cran (@1.20)                     
#>  lattice       0.20-35    2017-03-25 CRAN (R 3.3.3)                   
#>  lwgeom        0.1-5      2018-03-20 Github (r-spatial/lwgeom@389090c)
#>  magrittr      1.5        2014-11-22 CRAN (R 3.3.2)                   
#>  maptools      0.9-2      2017-03-25 cran (@0.9-2)                    
#>  memoise       1.1.0      2017-04-21 CRAN (R 3.4.1)                   
#>  methods     * 3.4.4      2018-03-16 local                            
#>  mime          0.5        2016-07-07 CRAN (R 3.3.2)                   
#>  openxlsx      4.0.17     2017-03-23 CRAN (R 3.4.1)                   
#>  pillar        1.2.1      2018-02-27 CRAN (R 3.4.4)                   
#>  pkgconfig     2.0.1      2017-03-21 cran (@2.0.1)                    
#>  R.methodsS3   1.7.1      2016-02-16 cran (@1.7.1)                    
#>  R.oo          1.21.0     2016-11-01 cran (@1.21.0)                   
#>  R.utils       2.6.0      2017-11-05 CRAN (R 3.4.2)                   
#>  R6            2.2.2      2017-06-17 cran (@2.2.2)                    
#>  raster        2.6-7      2017-11-13 CRAN (R 3.4.2)                   
#>  Rcpp          0.12.16    2018-03-13 CRAN (R 3.4.4)                   
#>  rgdal         1.2-18     2018-03-17 CRAN (R 3.4.4)                   
#>  rgeos         0.3-26     2017-10-31 cran (@0.3-26)                   
#>  rlang         0.2.0.9001 2018-04-10 Github (r-lib/rlang@70d2d40)     
#>  rmarkdown     1.9        2018-03-01 CRAN (R 3.4.4)                   
#>  rprojroot     1.3-2      2018-01-03 CRAN (R 3.4.3)                   
#>  sf          * 0.6-2      2018-04-09 Github (r-spatial/sf@b7de234)    
#>  sp            1.2-7      2018-01-19 cran (@1.2-7)                    
#>  stats       * 3.4.4      2018-03-16 local                            
#>  stplanr     * 0.2.3.9000 2018-04-10 Github (ropensci/stplanr@4b8abda)
#>  stringi       1.1.7      2018-03-12 CRAN (R 3.4.4)                   
#>  stringr       1.3.0      2018-02-19 cran (@1.3.0)                    
#>  tibble        1.4.2      2018-01-22 cran (@1.4.2)                    
#>  tools         3.4.4      2018-03-16 local                            
#>  udunits2      0.13       2016-11-17 cran (@0.13)                     
#>  units         0.5-1      2018-01-08 cran (@0.5-1)                    
#>  utils       * 3.4.4      2018-03-16 local                            
#>  withr         2.1.2      2018-04-07 Github (jimhester/withr@79d7b0d) 
#>  xml2          1.2.0      2018-01-24 CRAN (R 3.4.3)                   
#>  yaml          2.1.18     2018-03-08 CRAN (R 3.4.4)

[Robinlovelace]

Heads-up @cmcaine this may be something you're interested in (discussed Friday last week).

[cmcaine]

Thanks :) I'll take a look tomorrow.

…

On Tue, 1 May 2018, 10:18 Robin, ***@***.***> wrote: Heads-up @cmcaine <https://github.com/cmcaine> this may be something you're interested in (discussed Friday last week). — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#249 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AFuQeV4KaTe8uES93SaoLjXTlA8K7a1pks5tuChkgaJpZM4TPD8A> .

[mem48]

The problem seems to be related to tidyverse/dplyr#3279 and is happending in lines 555 to 571

sldf <- dplyr::bind_rows(
        lapply(
          1:length(routecoords),
          function(x){
            as.data.frame(routecoords[[x]]) %>%
              dplyr::mutate(linenum = x)
          })
      ) %>%
      sf::st_as_sf(
        coords = utils::head(colnames(.),-2),
        crs = sf::st_crs(sln@sl)$epsg
      ) %>%
      dplyr::group_by(.data$linenum) %>%
      dplyr::summarise() %>%
      sf::st_cast("LINESTRING") %>%
      dplyr::bind_cols(routedata) %>%
      dplyr::select(-.data$linenum)

The solution would be to go straight from the dataframe of coordiantes to a matrix of cooridnates and then a LINESTRING rather than goin via POINTS

[Robinlovelace]

On a related note, this also fails:

SLN_sf2 = SpatialLinesNetwork(routes_fast_sf) 
shortpath <- sum_network_routes(SLN_sf2, 1, 2, sumvars = "length") 

[richardellison]

Sorry for being slow on this! Definitely has to do with the issue that @mem48 refers to. I will need to fix it.

[Robinlovelace]

Hi @richardellison thanks for confirming. As discussed with @mem48 (who works with me here in Leeds) I've given it a good go but have failed to fix it so far. I suspect it's a regression silently introduced by dplyr that I only discovered with the above example. Any ideas towards a solution? Can give it another go but suspect, as the author of the function, you're better qualified to succeed than me!

[richardellison]

Adding do_union to summarise() appears to have fixed this. I don't think this is a regression, just that I must never have tested it on routes that did not have only increasing x-coordinates.