我试图在用户点击操作按钮后使用滑动条值进行闪亮计算。这些滑动条代表五个维度,用户可以拖动这些维度来设置每个变量的首选权重值。单击操作按钮后,应用程序应显示荷兰地图以及基于这些权重值具有最高“得分”的城市。目前我已经运行了以下代码,但是在您单击“操作”按钮后非常缓慢,因为它会在反应性滑动条更改时自动重新执行。是否有人建议让我的代码更快,并且操作按钮eventReactive以便必须单击它以让闪亮知道进行新的计算?
我已经在主要脚本中添加了一个gist链接,因为我想不出更简单的方式来共享我的数据(spatialpolygonsdataframe):https://gist.github.com/anonymous/602948bce91e61f36fd3eb0c4259d26c
Shiny app的屏幕截图:https://i.stack.imgur.com/k0Vg9.png
app.R
library(shiny)
library(leaflet)
ui <- fluidPage(
titlePanel("Quality-of-life-o-meter of The Netherlands"),
sidebarLayout(
sidebarPanel(
h3("Dimensions"),
h6("Assign a weight value for every dimension and press the 'find!' button"),
sliderInput("housingslider",
label = h4("Housing"),
min = 1,
max = 5,
value = 1),
sliderInput("populationslider",
label = h4("Population"),
min = 1,
max = 5,
value = 1),
sliderInput("provisionsslider",
label = h4("Provisions"),
min = 1,
max = 5,
value = 1),
sliderInput("safetyslider",
label = h4("Safety"),
min = 1,
max = 5,
value = 1),
sliderInput("physicalenvslider",
label = h4("Physical Environment"),
min = 1,
max = 5,
value = 1),
actionButton("action", label = "Find!")
),
mainPanel(
leafletOutput("mymap", height = "800")
)
)
)
server <- function(input, output, session) {
output$mymap <-renderLeaflet({
# Create interactive map of Total Score of Muncipalities in 2016 to display first
leaflet(data = MunScores2016) %>% addTiles() %>%
addPolygons(fillColor = ~pal(Total_Score_2016),
fillOpacity = 1,
color = 'white',
weight = 1,
popup = popup_dat) %>%
addLegend("bottomright", # Legend position
pal=pal, # color palette
values=~Total_Score_2016, # legend values
opacity = 0.7,
title="Percentage difference from national average")
})
observeEvent(input$action,
output$mymap <-renderLeaflet({
Total_Score <- NA
Total_Score <- ((input$housingslider * MunScores2016$Housing_Score_2016 +
input$populationslider * MunScores2016$Population_Score_2016 +
input$provisionsslider * MunScores2016$Provisions_Score_2016 +
input$safetyslider * MunScores2016$Safety_Score_2016 +
input$physicalenvslider * MunScores2016$PhysicalEnvironment_Score_2016)/
(input$housingslider + input$populationslider + input$provisionsslider + input$safetyslider + input$physicalenvslider))
#Create interactive map
leaflet(data = MunScores2016) %>% addTiles() %>%
addPolygons(fillColor = ~pal(Total_Score),
fillOpacity = 1,
color = 'white',
weight = 1,
popup = paste0("<strong>Municipality:</strong>",
MunScores2016$Municipality_Name,
"<br><strong>Quality-of-life-o-meter says: </strong>",
Total_Score)) %>%
addLegend("bottomright", # Legend position
pal=pal, # color palette
values=~Total_Score_2016, # legend values
opacity = 0.7,
title="Weighted percentage difference from national average")
})
)
}
shinyApp(ui = ui, server = server)
答案 0 :(得分:1)
您的代码很慢,因为您在不知情的情况下不断重新创建所有内容。
让我先解决滑块刷新问题:
在observeEvent
内(实际上只有在按下按钮时才触发),您将反应环境分配给output$mymap
(所有渲染功能都是反应环境)。 / p>
在 renderLeaflet
内,然后命名输入参数,Shiny将自动跟踪这些输入作为可观察对象
因此,一旦单击按钮,渲染函数的赋值就会发生,但渲染函数本身不仅限于此事件,因为反应环境比分配电话更长寿。
有两种方法可以将此逻辑从渲染函数中拉出来。
1)在每个输入参考周围使用isolate
。 (即isolate(input$housingslider)
)。这样你抑制到输入滑块的反应链接,渲染可观察的将不再有更新触发器,只在你创建时更新一次。
output$mymap <-renderLeaflet({
Total_Score <- NA
# isolate(expr) prevents observables in expr from being subscribed to.
Total_Score <- isolate(
(
input$housingslider * MunScores2016$Housing_Score_2016
+ input$populationslider * MunScores2016$Population_Score_2016
+ input$provisionsslider * MunScores2016$Provisions_Score_2016
+ input$safetyslider * MunScores2016$Safety_Score_2016
+ input$physicalenvslider * PhysicalEnvironment_Score_2016
) / (
input$housingslider
+ input$populationslider
+ input$provisionsslider
+ input$safetyslider
+ input$physicalenvslider
)
)
...
})
2)更好:将Total_Score
的计算移到renderLeaflet
函数的之外。这样,输入的可观察对象在被动环境中没有命名,即renderLeaflet
,它只知道它无法订阅的值Total_Score
。 。输入的observable只属于外部observeEvent
,它将忽略除操作按钮点击之外的输入更改。
observeEvent(input$action, {
Total_Score <- NA
# isolate(expr) prevents observables in expr from being subscribed to.
Total_Score <- (
input$housingslider * MunScores2016$Housing_Score_2016
+ input$populationslider * MunScores2016$Population_Score_2016
+ input$provisionsslider * MunScores2016$Provisions_Score_2016
+ input$safetyslider * MunScores2016$Safety_Score_2016
+ input$physicalenvslider * PhysicalEnvironment_Score_2016
) / (
input$housingslider
+ input$populationslider
+ input$provisionsslider
+ input$safetyslider
+ input$physicalenvslider
)
output$mymap <-renderLeaflet({
# Use Total_Score here and no input$
...
})
})
但是现在你实际应该实现这个:使用leafletProxy
!!
每次调用renderLeaflet
时,都会销毁之前使用过的地图对象,并创建一个全新的地图对象(并再次创建所有图块,这也意味着获取这些资源)。只要看到每次都重置缩放和查看位置,您就会意识到这一点。
有一种方法可以重复使用之前创建的地图。
函数leafletProxy(<name>)
可让您访问使用output$<name> <- renderLeaflet(...)
创建的地图,并可访问常规传单调用的所有功能。
因此,我们将代码重写为在开始时只有一个渲染函数,而不再进行重新创建。服务器功能如下:
server <- function(input, output, session) {
# Create Map once.
output$mymap <-renderLeaflet({
leaflet(data = MunScores2016)
%>% addTiles()
%>% addPolygons(
fillColor = ~pal(Total_Score_2016),
fill2Opacity = 1,
color = 'white',
weight = 1,
popup = popup_dat)
%>% addLegend(
"bottomright",
pal=pal,
values=~Total_Score_2016,
opacity = 0.7,
title="Percentage difference from national average")
})
observeEvent(input$action, {
Total_Score <- NA
Total_Score <- ((input$housingslider * MunScores2016$Housing_Score_2016 +
input$populationslider * MunScores2016$Population_Score_2016 +
input$provisionsslider * MunScores2016$Provisions_Score_2016 +
input$safetyslider * MunScores2016$Safety_Score_2016 +
input$physicalenvslider * MunScores2016$PhysicalEnvironment_Score_2016)/
(input$housingslider + input$populationslider + input$provisionsslider + input$safetyslider + input$physicalenvslider))
# Just update the existing map
leafletProxy("mymap")
# This you have to do now: remove the existing polygons, before you paint new ones.
%>% clearShapes()
%>% addPolygons(
fillColor = ~pal(Total_Score),
fillOpacity = 1,
color = 'white',
weight = 1,
popup = paste0("<strong>Municipality:</strong>",
MunScores2016$Municipality_Name,
"<br><strong>Quality-of-life-o-meter says: </strong>",
Total_Score))
# As you can see, no addLegend is needed, since the legend does not change.
})
}
使用此功能,上述所有刷新问题都不再发生,因为您不会为地图重新创建反应环境。不过,我认为第一部分将来会有所帮助。