server.R

#
# This is the server logic of a Shiny web application. You can run the 
# application by clicking 'Run App' above.
#
# Find out more about building applications with Shiny here:
# 
#    http://shiny.rstudio.com/
#
source('helper.R')

HazLevels <- data.frame(ystart = c(0,50,100,150,200,300),
                        yend = c(50,100,150,200,300,500),
                        HazColors = c('green','yellow','orange','red','purple','maroon'))

# Define server logic required to draw a histogram
shinyServer(function(input, output, session) {
  isolate({
    withProgress({
      setProgress(message = "Loading...")
    })})
  
  output$correlationplot = renderPlot({
    
    pollutantsonly = df_CurrentCity() %>%
      select('CO','O3','NO2','SO2') %>%
      filter(complete.cases(.))
    
    corrplot(cor(pollutantsonly),
             order = "hclust",
             title = "Correlation of Pollutant Levels",
             mar=c(0,0,1,0))
  })
  
  possibleStates <- reactive({
    stateList = df %>%
      filter(measurementyear == input$yearselected) %>%
      select(State) %>%
      unique() %>%
      arrange(State) %>%
      rbind(data.frame(State = 'ALL'),.)
    stateList$State
  })
  
  observe({
    updateSelectizeInput(session, "stateselected",
                         choices = possibleStates(),
                         selected = possibleStates()[1],
                         server = TRUE)
  })
  
  AQI_byPollutantYear <- reactive({
    AnnualCityAQI %>% 
      filter(PollutantType == input$pollutantselected &
               measurementyear == input$yearselected)
  })
  
  df_CurrentCity <- reactive({
    df %>% filter(input$cityselected == City_State) %>%
      select(City, Date.Local, NO2, O3, SO2, CO, measurementyear, month, City_State, lon, lat)
  })
  
  output$NumModerateDays_NO2 <- renderValueBox({
    daysOverCurrentCity = df_CurrentCity() %>% filter(., NO2 >= 51) %>% count()
    daysAvailableCurrentCity = df_CurrentCity() %>% count()
    
    valueBox(round(365 * daysOverCurrentCity[[1]] / daysAvailableCurrentCity[[1]]),
             "Days of Moderate NO2 Level Per Year", color = "yellow")
  })
  
  output$NumModerateDays_O3 <- renderValueBox({
    daysOverCurrentCity = df_CurrentCity() %>% filter(., O3 >= 51) %>% count()
    daysAvailableCurrentCity = df_CurrentCity() %>% count()
    
    valueBox(round(365 * daysOverCurrentCity[[1]] / daysAvailableCurrentCity[[1]]),
             "Days of Moderate O3 Level Per Year", color = "yellow")
  })
  
  output$NumModerateDays_CO <- renderValueBox({
    daysOverCurrentCity = df_CurrentCity() %>% filter(., CO >= 51) %>% count()
    daysAvailableCurrentCity = df_CurrentCity() %>% count()
    
    valueBox(round(365 * daysOverCurrentCity[[1]] / daysAvailableCurrentCity[[1]]),
             "Days of Moderate CO Level Per Year", color = "yellow")
  })
  output$NumModerateDays_SO2 <- renderValueBox({
    daysOverCurrentCity = df_CurrentCity() %>% filter(., SO2 >= 51) %>% count()
    daysAvailableCurrentCity = df_CurrentCity() %>% count()
    
    valueBox(round(365 * daysOverCurrentCity[[1]] / daysAvailableCurrentCity[[1]]),
             "Days of Moderate SO2 Level Per Year", color = "yellow")
  })
  
  output$NumHighDays_NO2 <- renderValueBox({
    daysOverCurrentCity = df_CurrentCity() %>% filter(., NO2 >= 101) %>% count()
    daysAvailableCurrentCity = df_CurrentCity() %>% count()
    
    valueBox(round(365 * daysOverCurrentCity[[1]] / daysAvailableCurrentCity[[1]]),
             "Days of High NO2 Level Per Year", color = "orange")
  })
  
  output$NumHighDays_O3 <- renderValueBox({
    daysOverCurrentCity = df_CurrentCity() %>% filter(., O3 >= 101) %>% count()
    daysAvailableCurrentCity = df_CurrentCity() %>% count()
    
    valueBox(round(365 * daysOverCurrentCity[[1]] / daysAvailableCurrentCity[[1]]),
             "Days of High O3 Level Per Year", color = "orange")
  })
  
  output$NumHighDays_CO <- renderValueBox({
    daysOverCurrentCity = df_CurrentCity() %>% filter(., CO >= 101) %>% count()
    daysAvailableCurrentCity = df_CurrentCity() %>% count()
    
    valueBox(round(365 * daysOverCurrentCity[[1]] / daysAvailableCurrentCity[[1]]),
             "Days of High CO Level Per Year", color = "orange")
  })
  output$NumHighDays_SO2 <- renderValueBox({
    daysOverCurrentCity = df_CurrentCity() %>% filter(., SO2 >= 101) %>% count()
    daysAvailableCurrentCity = df_CurrentCity() %>% count()
    
    valueBox(round(365 * daysOverCurrentCity[[1]] / daysAvailableCurrentCity[[1]]),
             "Days of High SO2 Level Per Year", color = "orange")
  })
  
  # Timeseries:
  output$timeplot <- renderPlotly({
    x <- list(
      title = "Date"
    )
    y <- list(
      title = "AQI"
    )
    
    
    plot_ly(df_CurrentCity()) %>%
      add_markers(x = ~Date.Local, y = ~NO2, name = "NO2") %>%
      add_markers(x = ~Date.Local, y = ~O3, name = "O3") %>%
      add_markers(x = ~Date.Local, y = ~CO, name = "CO") %>%
      add_markers(x = ~Date.Local, y = ~SO2, name = "SO2") %>%
      #add_lines(x = c(-Inf, Inf), y = c(50, 50)) %>%
      #add_lines(x = c(-Inf, Inf), y = c(100, 100)) %>% 
      layout(xaxis = x, yaxis = y, title = paste("Air Pollution History for", input$cityselected))
  })

  output$popscatter = renderPlot({
    currstateonly = AQI_byPollutantYear() %>% filter(state == input$stateselected)
    
    ggplot(AQI_byPollutantYear()) +
      geom_point(aes(x = citypop, y = AQI)) +
      scale_x_log10() +
      geom_smooth(aes(x = citypop, y = AQI),method = 'lm') +
      geom_point(data = currstateonly, aes(x = citypop, y = AQI), color = 'blue', size = 6) +
      ggtitle('Effect of City Size on Pollution', subtitle = '(Selected State in Blue)') + 
      xlab('City Population') +
      ylab(paste("Air Quality Index for",input$pollutantselected)) +
      theme_bw(base_size = 18)
  })
  
  # Boxplots comparing different cities in the same state:
  output$citycompare <- renderPlot({
    
    if (input$stateselected != "ALL") {
      
      newdf = df %>%
        filter(measurementyear == input$yearselected & State == input$stateselected) #%>%
        #mutate(City, pollutantOfInterest = input$pollutantselected)
      
      max_y = newdf %>% select(input$pollutantselected) %>% max()
      
      ggplot(newdf) +
        geom_boxplot(data = newdf,
                     aes_string(x = paste0("reorder(City, ",input$pollutantselected,")"),
                                y = input$pollutantselected)) +
        geom_rect(data = HazLevels,
                  aes(xmin = -Inf, xmax = Inf, ymin = ystart, ymax = yend),
                  fill = HazLevels$HazColors,
                  alpha = 0.3) + 
        labs(x = "City") +
        theme_minimal(base_size = 18) + 
        coord_cartesian(expand = TRUE, ylim = c(0, max_y)) + 
        ggtitle("Pollution Levels By City")
    } else {
      
      newdf = df %>% filter(measurementyear == input$yearselected)
      
      bestcities = newdf[,c("City_State", input$pollutantselected)] %>%
        group_by(City_State) %>% 
        summarise(median_ = median(get(input$pollutantselected))) %>%
        arrange(median_) %>% top_n(-10)
      
      #print(head(bestcities))
      
      topten_df = newdf %>%
        filter(newdf$City_State %in% bestcities$City_State)
      
      max_y = topten_df %>% select(input$pollutantselected) %>% max()
      
      ggplot(topten_df) +
        geom_boxplot(aes_string(x = paste0("reorder(City, ",input$pollutantselected,")"),
                                y = input$pollutantselected)) +
        geom_rect(data = HazLevels,
                  aes(xmin = -Inf, xmax = Inf,
                      ymin = ystart, ymax = yend),
                  fill = HazLevels$HazColors,
                  alpha = 0.3) + 
        labs(x = "City") +
        theme_minimal(base_size = 18) + 
        coord_cartesian(expand = TRUE, ylim = c(0, max_y)) + 
        ggtitle("Least Polluted Cities")
    }
  })
  
  # Map showing pollution bubbles
  output$pollutionmap <- renderLeaflet({
    
    AQI_byPollutantYear = AQI_byPollutantYear() %>%
      filter(ifelse(input$stateselected == 'ALL', T, state == input$stateselected))
    
    pal <- colorNumeric(palette = "Reds", domain = AQI_byPollutantYear$AQI)
    
    AQI_byPollutantYear %>%
      leaflet() %>% addProviderTiles('OpenTopoMap', group = "Topo") %>%
      addCircleMarkers(lng = AQI_byPollutantYear$long, lat = AQI_byPollutantYear$lat,
                       popup = paste(AQI_byPollutantYear$City_State, '<br>',
                                   'Population:', AQI_byPollutantYear$citypop, '<br>',
                                   'Average Annual AQI:', round(AQI_byPollutantYear$AQI)),
                       fillColor = 'Blue',
                       #opacity = AQI_byPollutantYear$AQI/max(AQI_byPollutantYear$AQI),
                       fillOpacity = AQI_byPollutantYear$AQI/max(AQI_byPollutantYear$AQI),
                       radius = 20 * sqrt(AQI_byPollutantYear$citypop/max(AQI_byPollutantYear$citypop, na.rm = T)),
                       color = 'Red',
                       stroke = F) #%>%
      #addLegend("bottomright", pal = pal, values = AQI_byPollutantYear$PollutantType, title = "AQI")
  })
})