Visualizations

• Draggable barplot with plotly
• Combined distribution of all experts’ elicitations

Automation

• Send e-mails using mailR
• Enter quantiles in specific order using rhandsontable

Deployment

• Save results in single folder per session with docker volume
• Authorization via ShinyProxy

Weather Forecast Toulouse

library(rhandsontable)
rhandsontable(data, readOnly = TRUE) %>%
        hot_col(col = 4, type = "numeric") %>%
        hot_cell(row = editRow() + 1, col = 4, readOnly = FALSE) %>%
        # To make sure enough digits are shown
        hot_cols(renderer = htmlwidgets::JS("safeHtmlRenderer")) %>%
        hot_cols(renderer = paste0("
                                function (instance, td, row, col) {
                                Handsontable.renderers.TextRenderer.apply(this, arguments);
                                var row_index = ", paste("[", paste(editRow(), collapse = ","), "]"), ";
                                for (i = 0; i < row_index.length; i++)
                                if (row == row_index[i] & col == 3) {
                                td.style.background = '", editColor(), "';
                                }
                                }"))

library(plotly)
plot_ly(data = plotData, 
                x = ~x, y = ~c(0, diff(y)/diff(x)),
                type = "scatter", mode = "lines",
                line = list(shape = "vh"), fill = "tozeroy") %>%
        # add vertical lines
        layout(shapes = lapply(plotData$x, function(xx)
                            list(type = "line", x0 = xx, x1 = xx, y0 = 0, y1 = 1, yref = "paper"))) %>%
        # allow to drag lines
        config(edits = list(shapePosition = TRUE)) 

With ShinyProxy
Username is passed from LDAP login to R Shiny App

login <- Sys.getenv("SHINYPROXY_USERNAME")

Create Docker Volume

• Single folder per session
• Not single rectangular data file
• Persistent data storage (remote)

Session