Skip to contents

Advanced Interactivity

Source: vignettes/advanced_interactivity.Rmd
advanced_interactivity.Rmd

This vignette demonstrates how to enhance the interactivity of gg1d visualizations by adding custom tooltips or cross-linking them with other plots, such as UMAP visualizations.

Adding custom tooltips

Custom tooltips make your visualizations more interactive and insightful. To add tooltips, create a new column with the _tooltip suffix that contains the desired text.

For example, lets add custom-tooltips to the Magpie column of the lazy_birdwatcher dataset by creating a Magpies_tooltip column. Values should represent what text to show in the tooltip.

# No Custom Tooltip
gg1d(lazy_birdwatcher)

We create a Magpies_tooltip character column and set the value to either "More than 3 magpies!' or "How boring. Too few magpies (nMagpies)". gg1d will automatically detect this column based on the name and set the tooltip accordingly.

# Add a custom tooltip column
lazy_birdwatcher["Magpies_tooltip"] <- ifelse(
  lazy_birdwatcher[["Magpies"]] > 3,
  yes = "More than 3 magpies!",
  no = paste0("How boring. Too few magpies (", lazy_birdwatcher[["Magpies"]], ")")
)

# Generate a gg1d plot with the custom tooltip
gg1d(lazy_birdwatcher)

Cross-linking gg1d plots with other visualisations

Cross-linking enables coordinated exploration of gg1d plots with other visualizations produced by custom scripts or other packages. This workflow demonstrates how to combine interactivity using ggiraph and patchwork.

As an example, lets create both a gg1d visualisation and a UMAP summarising the palmer_penguins dataset, then cross-link them so they can be coexplored.

Workflow

  1. Create your gg1d plot. Ensure you set col_id to embed a data-identifier required for cross-linking plots (in our case, we will use the unique name given to the penguins). Also set interactive = FALSE to force gg1 to returns a ggplot object instead of a htmlwidget (we’ll add the interactivity back in after we compose all our plots). For example, lets

  2. Create the ggplot you want to co-explore with, adding latent interactivity using the ggiraph package (e.g. use geom_point_interactive instead of geom_point and add a data_id aesthetic set to the same column as col_id)

  3. Compose the two ggplots together using patchwork

  4. Make the plots interactive using the ggiraph package

Example Code 

library(uwot)
library(ggplot2)
library(patchwork)
library(ggiraph)


# Prepare Data & Colour Schemes
penguins <- na.omit(read.csv(system.file("penguins.csv", package = "gg1d")))
palette_species <- c(Chinstrap = "#C55BCC", Adelie = "#FF7F02", Gentoo = "#047476")

# Create gg1d plot
gg1d_penguins <- gg1d(
  penguins,
  col_id = "name",
  interactive = FALSE,
  col_sort = "species",
  palettes = list(species = palette_species),
  options = gg1d_options(relative_height_numeric = 1.2, show_legend = FALSE)
)

# Perform the umap
umap_mx <- umap(penguins, scale = "scale", n_neighbors = 70)
df_umap <- as.data.frame(umap_mx)
colnames(df_umap) <- c("UMAP_1", "UMAP_2")
df_umap[["name"]] <- penguins[["name"]]
df_umap[["species"]] <- penguins[["species"]]

# Create the umap plot
gg_umap <- ggplot(df_umap, aes(x = UMAP_1, y = UMAP_2, colour = species, data_id = name)) +
  geom_point_interactive() +
  scale_color_manual(values = palette_species) +
  theme_bw()

# Combine the gg1d plot and the umap using patchwork
combined_plots <- free(gg_umap, type = "label") / free(gg1d_penguins, side = "tb")

# Make interactive
interactive_multiplot <- girafe(ggobj = combined_plots, height_svg = 6, width_svg = 10, options)

# Add some settings to choose how to make combined plots interactive
interactive_multiplot <- girafe_options(
  x = interactive_multiplot,
  opts_selection(type = "multiple", only_shiny = FALSE, css = "opacity: 1"),
  opts_selection_inv(css = "opacity: 0.12")
)

# Draw Interactive Multiplot
interactive_multiplot