suburbia.R

library(tidyverse)
library(sf)
library(sfnetworks)
library(igraph)
library(tidygraph)
library(ggspatial)
eds <- st_read("D:/Downloads/israel-ISR_gpkg/tel_aviv-4309.gpkg",layer = "edges") %>% st_transform(2039) %>% select(geom)
net <- as_sfnetwork(eds) %>% 
  convert(to_spatial_smooth)

cmps <- net %>% 
  components() %>% 
  `$`(membership)
fnet <- net %>% 
  mutate(cmps =cmps) %>% 
  filter(cmps == 1) %>% 
  convert(to_spatial_smooth)

matr <- fnet %>% 
  st_network_cost()
gc()
n <- nrow(matr)
hunds <- numeric(n)
maxs <- numeric(n)
for(i in 1:n){
  sorted = sort(matr[i,])
  hunds[i] <- sorted[100]
  maxs[i] <- sorted[n]
  print(i)
  if(i %% 1000==0) {gc()}
}
# sorted <- t(apply(matr,1,sort))
min100 <- min(hunds)
max100 <- max(hunds)
minmax <- min(maxs)
maxmax <- max(maxs)
mat100 <- (hunds - min100)/(max100 - min100)
matmax <- (maxs - minmax)/(maxmax-minmax)
hist(matmax)
med_max <- matmax %>% median()
hist(mat100)
med_hunds <- mat100 %>% median()
classes <- data.frame(x = mat100,y = matmax) %>% 
  mutate(close = matmax < med_max,
         urban = mat100 < med_hunds,
         clas = ifelse(close & urban, "core",
                       ifelse(close & !urban, "suburb",
                              ifelse(!close & urban,"exburb","far"))))
classes %>% ggplot(aes(x,y)) + geom_point()
p1 <- fnet %>% 
  mutate(clas = classes$clas) %>% 
  st_as_sf() %>% 
  ggplot(aes( color = clas)) + 
  annotation_map_tile(zoom = 10) + 
  geom_sf()
p1
fnet %>% 
  mutate(clas = classes$clas) %>% 
  st_as_sf() %>% 
  st_write("tlv_met_class.gpkg")