Leaflet Shiny 集成缓慢答案

【问题标题】：Leaflet Shiny Integration slowLeaflet Shiny 集成缓慢
【发布时间】：2017-11-05 11:25:50
【问题描述】：

我正在尝试使用传单在 Shiny 中构建交互式 Choropleth。但是，加载时间和重新创建时间真的很慢。任何加快速度的方法。

这里是整个应用程序文件夹的链接以及数据： https://www.dropbox.com/home/Leaflet_Shiny_app

全球.R

library(shinydashboard)
library(tidyverse)
library(ggvis)
library(leaflet)
library(WDI)
library(sp)

ui.R

header <- dashboardHeader(
  title = "Greenhouse gas (GHG) emissions"
  )

## Sidebar content
sidebar <- dashboardSidebar(
  sidebarMenu(
    menuItem("Interactive Choropleth", tabName = "choropleth")
  )
)  

## Body content
body <- dashboardBody(

  # First tab content
  tabItem("choropleth",

    fluidRow(
      column(width = 9,
        box(width = NULL, solidHeader = TRUE,
            title = "Greenhouse gas emissions (kt of CO2 equivalent)",
            leafletOutput("choropleth_ghg", height = 500)
        )
      ),
      column(width = 3,
        box(width = NULL, status = "warning",
          selectInput("year", "Year", 
                        choices = seq(1970, 2012, 1), 
                        selected = 2012)
        )
      )
    )
  )
)

dashboardPage(
  header,
  sidebar,
  body
)

服务器.R

# Read the dataset for choropleth
# From http://data.okfn.org/data/core/geo-countries#data
countries <- geojsonio::geojson_read("json/countries.geojson", what = "sp")

# Download the requested data by using the World Bank's API, 
# parse the resulting JSON file, and format it in long country-year format.
load("who_ghg.RData")

function(input, output, session) {

  # Interactive Choropleth map.........................................................

  # Reactive expression for the data subsetted to what the user selected
  countries_plus_ghg <- reactive({

    # Filter the data to select for the year user selected
    who_ghg_subset <- filter(who_ghg, year == input$year)

    # Merge a Spatial object having a data.frame for Choropleth map
    sp::merge(countries, who_ghg_subset, 
              by.x = "ISO_A3", by.y = "iso3c")
  })

  # Create the map
  output$choropleth_ghg <- renderLeaflet({
    leaflet(countries) %>% 
      setView(0, 20, zoom = 1) %>%
      addTiles()
  })

  # Observer to change the color of countries, labels and legends
  # based on the year user selects in the UI
  observe({
    dat <- countries_plus_ghg()

    # Define numeric vector bins to add some color
    bins <- ggplot2:::breaks(c(min(dat$EN.ATM.GHGT.KT.CE, na.rm = TRUE)
                               ,max(dat$EN.ATM.GHGT.KT.CE, na.rm = TRUE)),
                             "width",n = 5)

    # Call colorBin to generate a palette function that maps the RColorBrewer 
    #"YlOrRd" colors to our bins.
    pal <- colorBin("YlOrRd", 
                    domain = dat$EN.ATM.GHGT.KT.CE, 
                    bins =  bins)

    # Generate the labels with some HTML
    labels <- sprintf(
      "<strong>%s</strong><br/>%g",
      dat$country, dat$EN.ATM.GHGT.KT.CE
    ) %>% lapply(htmltools::HTML)

    leafletProxy("choropleth_ghg", data = dat) %>% 
      addPolygons(
        fillColor = ~pal(EN.ATM.GHGT.KT.CE),
        weight = 1,
        opacity = 1,
        color = "white",
        fillOpacity = 0.7,
        highlight = highlightOptions(
          weight = 2,
          color = "#666",
          dashArray = "",
          fillOpacity = 0.7,
          bringToFront = TRUE),
        label = labels,
        labelOptions = labelOptions(
          style = list("font-weight" = "normal", padding = "3px 8px"),
          textsize = "15px",
          direction = "auto")) %>% 
      clearControls() %>% 
      addLegend(pal = pal, values = ~EN.ATM.GHGT.KT.CE, opacity = 0.7, title = NULL,
                position = "bottomleft")
  })

}

【问题讨论】：

每次用户更改选择时，您都会下载数据。在启动时做更多的预处理（比如下载数据）。
我在一开始就下载了所有年份的数据，并下载了用户选择作为反应表达式的年份的子集。那也没有帮助。还有其他建议吗？
在另一个脚本中下载它并将其保存到 .Rdata 文件中。然后为您的程序加载该文件一次，但在反应之外的开头。根据需要在响应式中加载数据帧的子集。真的只是常识。这应该会让事情变得更快。
您好，@MikeWise 我按照建议更新了我的代码，但是，该应用程序的运行速度仍然很慢。你能看看吗？
你好，@yeedle 我按照建议更新了我的代码，但是，Shiny 应用程序仍然运行缓慢。你能看看吗？

标签： r leaflet shiny shinydashboard

【解决方案1】：

使用 rmapshaper::ms_simplify 简化几何图形有助于加快速度。

这就是我所做的-

# Topologically-aware geometry simplification using rmapshaper package, 
# keep = proportion of points to retain
countries_simple <- rmapshaper::ms_simplify(countries, keep = 0.05, keep_shapes = TRUE)

然后我在代码中使用了 countries_simple 而不是国家/地区。

【讨论】：