Help in generate map in Shiny

shiny
1

Friends could help me resolve the following issue:
I am inserting two executable codes below, the first generates a map, showing the route between two locations. In this case, the two locations to generate the map were defined: from = c(df_spec_clust[1, c("Longitude")], df_spec_clust[1, c("Latitude")])
and to = c (df_spec_prop [4, c ("Longitude")], df_spec_prop [4, c ("Latitude")])]. In the second, I would like to generate the map in Shiny format, but without defining the locations exactly as I did in the first code. I would like them to be selected from the filters I created (Filter 1 and Filter 2). However, I am unable to generate the map. Could you help me ?

First code

library(sf)
library(sfnetworks)
library(tmap)
library(rdist)
library(geosphere)

#database df
df <- structure(
  list(Property = c(1,2,3,4,5,6,7), Latitude = c(-24.779225, -24.789635, -24.763461, -24.794394, -24.747102,-24.781307,-24.761081),
    Longitude = c(-49.934816, -49.922324, -49.911616, -49.906262, -49.890796,-49.8875254,-49.8875254), 
    Waste = c(526, 350, 526, 469, 285, 433, 456)),class = "data.frame", row.names = c(NA, -7L))

#clusters
coordinates<-df[c("Latitude","Longitude")]
d<-as.dist(distm(coordinates[,2:1]))
fit.average<-hclust(d,method="average") 
k=3
clusters<-cutree(fit.average, k) 
nclusters<-matrix(table(clusters))  
df$cluster <- clusters 

#Create database df1
center<-matrix(nrow=k,ncol=2)
for(i in 1:k){
  center[i,]<-c(weighted.mean(subset(df,cluster==i)$Latitude,subset(df,cluster==i)$Waste),
                     weighted.mean(subset(df,cluster==i)$Longitude,subset(df,cluster==i)$Waste))}
coordinates$cluster<-clusters 
center<-cbind(center,matrix(c(1:k),ncol=1)) 
df1<-as.data.frame(center)
colnames(df1) <-c("Latitude", "Longitude", "cluster")

#specific cluster and specific property
df_spec_clust <- df1[df1$cluster,]
df_spec_prop<-df[df$Property,]

#create map
download.file("https://github.com/JovaniSouza/JovaniSouza5/raw/master/Test.zip", "Test.zip")
unzip("Test.zip")
roads = st_read("Test/regionbrazil.shp", quiet = TRUE) %>% 
  st_cast("LINESTRING")

# build sfnetwork
net = as_sfnetwork(roads, directed = FALSE) %>%
  activate("edges") %>%
  dplyr::mutate(weight = edge_length())

# routing
from = c(df_spec_clust[1, c("Longitude")], df_spec_clust[1, c("Latitude")])
to = c(df_spec_prop[4, c("Longitude")], df_spec_prop[4, c("Latitude")])
p1 = st_as_sf(data.frame(x = from[1], y = from[2]), coords = c("x", "y"), crs = st_crs(net))
p2 = st_as_sf(data.frame(x = to[1], y = to[2]), coords = c("x", "y"), crs = st_crs(net))
r = tidygraph::convert(net, to_spatial_shortest_paths, p1, p2)

# Extract the bbox for r 
bbox_r = st_as_sfc(r %>% activate(edges) %>% st_bbox())


# filter the net
small_net = st_filter(net, bbox_r)

# plot
plot1<-tm_shape(small_net %>% activate(edges) %>% st_as_sf()) +
  tm_lines() + 
  tm_shape(rbind(p1, p2)) + 
  tm_dots(col = "red", size = 0.5) + 
  tm_shape(r %>% activate(edges) %>% st_as_sf()) + 
  tm_lines(col = "red", lwd = 3)
plot1

Map generated by the code above

image
image616×599 1.96 KB

Second code

library(shiny)
library(rdist)
library(geosphere)
library(shinythemes)
library(sf)
library(tidygraph)
library(sfnetworks)
library(tmap)

#for the roads file
 download.file("https://github.com/JovaniSouza/JovaniSouza5/raw/master/Test.zip", "Test.zip")
 unzip("Test.zip")

function.cl<-function(df,k,Filter1,Filter2){
  
  #database df
  df <- structure(
    list(Property = c(1,2,3,4,5,6,7), Latitude = c(-24.779225, -24.789635, -24.763461, -24.794394, -24.747102,-24.781307,-24.761081),
         Longitude = c(-49.934816, -49.922324, -49.911616, -49.906262, -49.890796,-49.8875254,-49.8875254), 
         Waste = c(526, 350, 526, 469, 285, 433, 456)),class = "data.frame", row.names = c(NA, -7L))
  
  #clusters
  coordinates<-df[c("Latitude","Longitude")]
  d<-as.dist(distm(coordinates[,2:1]))
  fit.average<-hclust(d,method="average") 
  clusters<-cutree(fit.average, k) 
  nclusters<-matrix(table(clusters))  
  df$cluster <- clusters 
  
  #Create database df1
  center<-matrix(nrow=k,ncol=2)
  for(i in 1:k){
    center[i,]<-c(weighted.mean(subset(df,cluster==i)$Latitude,subset(df,cluster==i)$Waste),
                  weighted.mean(subset(df,cluster==i)$Longitude,subset(df,cluster==i)$Waste))}
  coordinates$cluster<-clusters 
  center<-cbind(center,matrix(c(1:k),ncol=1)) 
  df1<-as.data.frame(center)
  colnames(df1) <-c("Latitude", "Longitude", "cluster")
 
  # specific cluster and specific property
  df_spec_clust <- df1[df1$cluster==Filter1,]
  df_spec_prop<-df[df$Property==Filter2,]
  
 
  #create map
 
  roads = st_read("Test/regionbrazil.shp", quiet = TRUE) %>% 
    st_cast("LINESTRING")
  
  # build sfnetwork
  net = as_sfnetwork(roads, directed = FALSE) %>%
    activate("edges") %>%
    dplyr::mutate(weight = edge_length())
  
  # routing
  from = c(df_spec_clust[1, c("Longitude")], df_spec_clust[1, c("Latitude")])
  to = c(df_spec_prop[4, c("Longitude")], df_spec_prop[4, c("Latitude")])
  p1 = st_as_sf(data.frame(x = from[1], y = from[2]), coords = c("x", "y"), crs = st_crs(net))
  p2 = st_as_sf(data.frame(x = to[1], y = to[2]), coords = c("x", "y"), crs = st_crs(net))
  r = tidygraph::convert(net, to_spatial_shortest_paths, p1, p2)
  
  # Extract the bbox for r 
  bbox_r = st_as_sfc(r %>% activate(edges) %>% st_bbox())
  
  
  # filter the net
  small_net = st_filter(net, bbox_r)
  
  # plot
  plot1<-tm_shape(small_net %>% activate(edges) %>% st_as_sf()) +
    tm_lines() + 
    tm_shape(rbind(p1, p2)) + 
    tm_dots(col = "red", size = 0.5) + 
    tm_shape(r %>% activate(edges) %>% st_as_sf()) + 
    tm_lines(col = "red", lwd = 3)

  return(list(
    "Plot1" = plot1,
    "Data" =  df
  ))
}

ui <- bootstrapPage(
  navbarPage(theme = shinytheme("flatly"), collapsible = TRUE,
             "Cl", 
          tabPanel("",
           sidebarLayout(
             sidebarPanel(
               sliderInput("Slider", h5(""),
                           min = 2, max = 4, value = 3),
               selectInput("Filter1", label = h4("Select just one cluster"),""),
               selectInput("Filter2",label=h4("Select the cluster property"),""),
             ),
             mainPanel(
               tabsetPanel(
                 tabPanel("Map", plotOutput("Map1"))))
           ))))

server <- function(input, output, session) {
  
  Modelcl<-reactive({
    function.cl(df,input$Slider,input$Filter1,input$Filter2)
  })
  

  output$Map1 <- renderPlot({
    Modelcl()[[1]]
  })
  
  observeEvent(input$Slider, {
    abc <- req(Modelcl()$Data)
    updateSelectInput(session,'Filter1',
                      choices=sort(unique(abc$cluster)))
  }) 
  
  observeEvent(input$Filter1,{
    abc <- req(Modelcl()$Data) %>% filter(cluster == as.numeric(input$Filter1))
    updateSelectInput(session,'Filter2',
                      choices=sort(unique(abc$Property)))
  }) 
  
  
}

shinyApp(ui = ui, server = server)

For test: Map using leaflet package (It works)

library(shiny)
library(rdist)
library(geosphere)
library(shinythemes)
library(leaflet)
library(tidygraph)

function.cl<-function(df,k,Filter1,Filter2){
  
  #database df
  df <- structure(
    list(Property = c(1,2,3,4,5,6,7), Latitude = c(-24.779225, -24.789635, -24.763461, -24.794394, -24.747102,-24.781307,-24.761081),
         Longitude = c(-49.934816, -49.922324, -49.911616, -49.906262, -49.890796,-49.8875254,-49.8875254), 
         Waste = c(526, 350, 526, 469, 285, 433, 456)),class = "data.frame", row.names = c(NA, -7L))
  
  #clusters
  coordinates<-df[c("Latitude","Longitude")]
  d<-as.dist(distm(coordinates[,2:1]))
  fit.average<-hclust(d,method="average") 
  clusters<-cutree(fit.average, k) 
  nclusters<-matrix(table(clusters))  
  df$cluster <- clusters 
  
  #Create database df1
  center<-matrix(nrow=k,ncol=2)
  for(i in 1:k){
    center[i,]<-c(weighted.mean(subset(df,cluster==i)$Latitude,subset(df,cluster==i)$Waste),
                  weighted.mean(subset(df,cluster==i)$Longitude,subset(df,cluster==i)$Waste))}
  coordinates$cluster<-clusters 
  center<-cbind(center,matrix(c(1:k),ncol=1)) 
  df1<-as.data.frame(center)
  colnames(df1) <-c("Latitude", "Longitude", "cluster")
  
  #specify cluster and specific cluster and specific propertie
  df_spec_clust <- df1[df1$cluster==Filter1,]
  df_spec_prop<-df[df$Property==Filter2,]
  
  
  #color for map
  ai_colors <-c("red","gray","blue","orange","green","beige","darkgreen","lightgreen", "lightred", "darkblue","lightblue",
                "purple","darkpurple","pink", "cadetblue","white","darkred", "lightgray","black")
  clust_colors <- ai_colors[df$cluster]
  icons <- awesomeIcons(
    icon = 'ios-close',
    iconColor = 'black',
    library = 'ion',
    markerColor =  clust_colors)
  
  # create icon for map
  leafIcons <- icons(
    iconUrl = ifelse(df1$cluster,
                     
                     "https://image.flaticon.com/icons/svg/542/542461.svg"
    ),
    iconWidth = 30, iconHeight = 40,
    iconAnchorX = 25, iconAnchorY = 12)
  
  html_legend <- "<img src='https://image.flaticon.com/icons/svg/542/542461.svg'>"
  
# create map
  if(nrow(df_spec_clust)>0){
    clust_colors <- ai_colors[df_spec_clust$cluster]
    icons <- awesomeIcons(
      icon = 'ios-close',
      iconColor = 'black',
      library = 'ion',
      markerColor =  clust_colors)
    
  m1<-leaflet(df_spec_clust) %>% addTiles() %>% 
    addMarkers(~Longitude, ~Latitude, icon = leafIcons) %>%
    addAwesomeMarkers(leaflet(df_spec_prop) %>% addTiles(), lat=~df_spec_prop$Latitude, lng = ~df_spec_prop$Longitude, icon= icons,label=~cluster)

  for(i in 1:nrow(df_spec_clust)){
    df_line <- rbind(df_spec_prop[,c("Latitude","Longitude")],
                     df_spec_clust[i,c("Latitude","Longitude")])
    m1 <- m1 %>%
      addPolylines(data = df_line,
                   lat=~Latitude,
                   lng = ~Longitude,
                   color="red")
  }
  plot1<-m1} else plot1 <- NULL

  return(list(
    "Plot1" = plot1,
    "Data"= df
  ))
}

ui <- bootstrapPage(
  navbarPage(theme = shinytheme("flatly"), collapsible = TRUE,
             "Cl", 
             tabPanel("",
                      sidebarLayout(
                        sidebarPanel(
                          sliderInput("Slider", h5(""),
                                      min = 2, max = 4, value = 3),
                          selectInput("Filter1", label = h4("Select just one cluster"),""),
                          selectInput("Filter2",label=h4("Select the cluster property"),""),
                        ),
                        mainPanel(
                          tabsetPanel(
                            tabPanel("Map", uiOutput("Map1"))))
                      ))))

server <- function(input, output, session) {
  
  Modelcl<-reactive({
    function.cl(df,input$Slider,input$Filter1,input$Filter2)
  })
  
  output$Map1 <- renderUI({ 
    if(input$Filter1!="") 
      leafletOutput("Leaf1",width = "95%", height = "600") })

  output$Leaf1 <- renderLeaflet({
    req(Modelcl())[[1]]
  })
  
  
  observeEvent(input$Slider, {
    abc <- req(Modelcl()$Data)
    updateSelectInput(session,'Filter1',
                      choices=sort(unique(abc$cluster)))
  }) 
  
  observeEvent(input$Filter1,{
    abc <- req(Modelcl()$Data) %>% filter(cluster == as.numeric(input$Filter1))
    updateSelectInput(session,'Filter2',
                      choices=sort(unique(abc$Property)))
  }) 
  
  
}

shinyApp(ui = ui, server = server)
2

your map is static, yet, you have it be downloaded (17mb) and read in , within your function, i.e. sensitive to any slider change . This is a waste of network resources and time, and your app performance will suffer. Please clarify if the map is intending to be a fixed asset , i.e. the same every time, or if you plan to vary it and haven't demonstrated that yet. If it is indeed to be fixed, then prep it and load it once.

3

@nirgrahamuk, thanks for the reply. I improved the question, could you take a look if it became more understandable ?? So, the roads file will always be the same, the only thing that will change is the cluster and the property that the user will select (filters 1 and 2 of the code). From there, the route between these locations will be shown in the roads file. I removed the roads file from the function, inserted it after loading the libraries.

4

Thanks Jovani,
thats better, but you can go further as both roads and net are further transformations of the zipped map data, and will be the same every time you want to run the function, so should be also lifted out of the function.

  1. There's a problem where you set up your function.cl to take a df param, but this is ignored by the function which uses a locally scoped df hardcoded in. Later when you call function.cl, you pass an object that doesnt exist 'df' in as the df param.
    Either you keep passing df into function.cl but set up df similarly to the zip map (i.e. in the correct scope) or else omit a df param, and just use the hardcoded local df to the function.

  2. There's a logic problem where you rely on the output of function.cl to determine your pickers possible values , while those pickers determine whats in the function. This needs to be untangled at least somewhat I think

5

Thank you very much for your reply @nirgrahamuk . I did not understand some points you commented in the answer above, however for testing I managed to generate the map correctly for the problem in question using another package (leaflet). I entered the code above for your verification. Sometimes it can help you to see something that I am not seeing for the other code. I still couldn't think of a way to make it work using the sfnetworks package. Any help is appreciated. Thank you again!

6

Hello @nirgrahamuk , I had a small advance yesterday with the problem in question. I'll insert the code in this answer for you to take a look at. Basically, I entered the clusters commands outside the function to initialize the number of clusters (k = 3). Any subsequent calls will be clustered from the function.cl. I don't know if this way is right but it was the way I found it. I made changes in selecInput and server too. However, it is giving some errors in some combinations, for example cluster 1 and property 4, cluster 2 and property 5, cluster 3 and property 6. It gives the following error: ! AnyNA (x) is not TRUE . If you can take a look at the code I would appreciate it. If you think a new question is needed, I will ask. Thanks again.

library(shiny)
library(rdist)
library(geosphere)
library(shinythemes)
library(leaflet)
library(sf)
library(tidygraph)
library(sfnetworks)
library(tmap)

#for the roads file
download.file("https://github.com/JovaniSouza/JovaniSouza5/raw/master/Test.zip", "Test.zip")
unzip("Test.zip")

#database df
df <- structure(
  list(Property = c(1,2,3,4,5,6,7), Latitude = c(-24.779225, -24.789635, -24.763461, -24.794394, -24.747102,-24.781307,-24.761081),
       Longitude = c(-49.934816, -49.922324, -49.911616, -49.906262, -49.890796,-49.8875254,-49.8875254), 
       Waste = c(526, 350, 526, 469, 285, 433, 456)),class = "data.frame", row.names = c(NA, -7L))


#clusters
coordinates<-df[c("Latitude","Longitude")]
d<-as.dist(distm(coordinates[,2:1]))
fit.average<-hclust(d,method="average") 
clusters<-cutree(fit.average, 3) 
nclusters<-matrix(table(clusters))  
df$cluster <- clusters 

function.cl<-function(df,k,Filter1,Filter2){
  
  coordinates<-df[c("Latitude","Longitude")]
  d<-as.dist(distm(coordinates[,2:1]))
  fit.average<-hclust(d,method="average") 
  clusters<-cutree(fit.average, k) 
  nclusters<-matrix(table(clusters))  
  df$cluster <- clusters 

  #Create database df1
  center<-matrix(nrow=k,ncol=2)
  for(i in 1:k){
    center[i,]<-c(weighted.mean(subset(df,cluster==i)$Latitude,subset(df,cluster==i)$Waste),
                  weighted.mean(subset(df,cluster==i)$Longitude,subset(df,cluster==i)$Waste))}
  coordinates$cluster<-clusters 
  center<-cbind(center,matrix(c(1:k),ncol=1)) 
  df1<-as.data.frame(center)
  colnames(df1) <-c("Latitude", "Longitude", "cluster")
  
  # specific cluster and specific property
  df_spec_clust <- df1[df1$cluster==Filter1,]
  df_spec_prop<-df[df$Property==Filter2,]
  
  
  #create map
  
  roads = st_read("Test/regionbrazil.shp", quiet = TRUE) %>% 
    st_cast("LINESTRING")
  
  # build sfnetwork
  net = as_sfnetwork(roads, directed = FALSE) %>%
    activate("edges") %>%
    dplyr::mutate(weight = edge_length())
  
  # routing
  from = c(df_spec_clust[1, c("Longitude")], df_spec_clust[1, c("Latitude")])
  to = c(df_spec_prop[1, c("Longitude")], df_spec_prop[1, c("Latitude")])
  p1 = st_as_sf(data.frame(x = from[1], y = from[2]), coords = c("x", "y"), crs = st_crs(net))
  p2 = st_as_sf(data.frame(x = to[1], y = to[2]), coords = c("x", "y"), crs = st_crs(net))
  r = tidygraph::convert(net, to_spatial_shortest_paths, p1, p2)
  
  # Extract the bbox for r 
  bbox_r = st_as_sfc(r %>% activate(edges) %>% st_bbox())
  
  
  # filter the net
  small_net = st_filter(net, bbox_r)
  
  # plot
  plot1<-tm_shape(small_net %>% activate(edges) %>% st_as_sf()) +
    tm_lines() + 
    tm_shape(rbind(p1, p2)) + 
    tm_dots(col = "red", size = 0.5) + 
    tm_shape(r %>% activate(edges) %>% st_as_sf()) + 
    tm_lines(col = "red", lwd = 3)
  
  return(list(
    "Plot1" = plot1,
    "Data" =  df
  ))
}

ui <- bootstrapPage(
  navbarPage(theme = shinytheme("flatly"), collapsible = TRUE,
             "Cl", 
             tabPanel("",
                      sidebarLayout(
                        sidebarPanel(
                          sliderInput("Slider", h5(""),
                                      min = 2, max = 4, value = 3),
                          selectInput("Filter1", label = h4("Select just one cluster"),
                                      choices=unique(df$cluster), selected=1),
                          selectInput("Filter2",label=h4("Select the cluster property"),
                                      choices=df$Property, selected=1)
                        ),
                        mainPanel(
                          tabsetPanel(
                            tabPanel("Map", plotOutput("Map1"))))
                      ))))

server <- function(input, output, session) {
  
  Modelcl<-reactive({
    req(input$Slider,input$Filter1,input$Filter2)
    function.cl(df,input$Slider,input$Filter1,input$Filter2)
  })
  
  output$Map1 <- renderPlot({
    Modelcl()[[1]]
  })
  
  observeEvent(input$Slider, {
    abc <- Modelcl()[[2]]
    updateSelectInput(session,'Filter1',
                      choices=sort(unique(abc$cluster)))
  }) 
  
  observeEvent(input$Filter1,{
    abc <- Modelcl()[[2]] %>% filter(cluster == as.numeric(input$Filter1))
    updateSelectInput(session,'Filter2',
                      choices=sort(unique(abc$Property)))
  }) 
  
}

shinyApp(ui = ui, server = server)
7

you should take out the roads and net object creation and put them in another script that runs once and puts net object into a file representation with saveRDS().
Then your main shiny app need only net<-readRDS("net.RDS")
and this will save you something like 20-30 seconds each time the function computes.

also recommend you put shinycssloaders withSpinner function on your plotOutput so that theres a good visual indicator of calculation happening.

Then you can use debug() or browser() to look at your function call in the failure conditions you found, such as when cluster 2 and property 5.

8

This topic was automatically closed 54 days after the last reply. New replies are no longer allowed.

If you have a query related to it or one of the replies, start a new topic and refer back with a link.