sparkline

[IgnatiusPang]

Dear Developers,

Just a quick question, would the gt package be compatible with the sparkline package (https://github.com/htmlwidgets/sparkline)? (e.g. show sparklines within tables created by gt.)

Kind Regards,

Igy

[rich-iannone]

The text_transform() function is currently a cumbersome way of getting any HTML you want in specific cells. So, it’s theoretically possible to get output from the sparkline package into specific cells of a gt table. A step in this direction is #155, which is specific to putting in ggplot plots in a column but, should we get the implementation right, this could be extended to output from the sparkline package.

So, we should consider creating a fmt_ function that wraps sparkline and makes it easy to add output to specified table cells.

[IgnatiusPang]

Agree that a fmt_ function that wraps sparkline and/or based on ggplot will be useful.

[aetiologicCanada]

I note this: https://leonawicz.github.io/HtmlWidgetExamples/ex_dt_sparkline.html with interest. It is a datatable approach, but may provide ideas to the gt team.

[jthomasmock]

Another anecdote for ideas to bother from for gt - here is a reprex example for embedding:

• plotly
• sparklines

via the reactable table package.

Code Snippet

library(reactable)
library(tidyverse)
library(plotly)
library(glue)
library(htmltools)
library(sparkline)

car_df <- mtcars %>%
  group_by(cyl) %>% 
  summarize(mpg = list(mpg)) %>% 
  dplyr::mutate(
    Sparkline = NA,
    color = c("red", "blue", "green")
  )

car_df %>% 
  reactable(
    # Define columns
    # hide two columns and create a sparkline from the nested MPG data
    columns = list(
      mpg = colDef(show = FALSE),
      color = colDef(show = FALSE),
      Sparkline = colDef(cell = function(value, index){
        sparkline(car_df$mpg[[index]])
      })
    ),
    
    # Details creates a dropdown area that we can fill with a div
    details = function(index){
      match_cyl <- car_df$cyl[index]
      plots_data <- filter(mtcars, cyl == match_cyl)
      plot_col <- filter(car_df, cyl == match_cyl) %>% pull(color)
      htmltools::div(style = "padding: 30px", {
        p1 <- ggplot(plots_data, aes(x = mpg, y = disp)) +
          geom_point(data = mtcars, color = "grey", alpha = 0.5) +
          geom_point(color = plot_col) +
          theme_minimal() +
          scale_x_continuous(limits = c(0, max(mtcars$mpg) * 1.1)) +
          scale_y_continuous(limits = c(0, max(mtcars$disp) * 1.1)) +
          geom_hline(yintercept = 0) +
          geom_vline(xintercept = 0) +
          labs(title = glue::glue("Plot for {match_cyl} cylinders."),
               x = "MPG", y = "Displacement")
        
        plotly::ggplotly(p1)
      }
      )
    }
  )

The kableExtra package has a different implementation and requires explicitly loading a "blank" sparkline(0) to load the JS libs ahead of time.

kableExtra Example

[novotny1akub]

@rich-iannone could you please elaborate on how to use text_transform() to get sparkline in gt? I tried to do that, but had no success. Here is a reprex on Stack Overflow. It seems to me that I have managed to get the sparkline code there, but I struggle with getting the sparkline dependencies and to overally get it working.

[rich-iannone]

I think this is now reasonably solved with #1431.