Skip to contents

Examples

Source: vignettes/Examples.Rmd
Examples.Rmd

Overview

sugarglider provides ggplot2 extensions to create glyph maps that visualize multivariate spatio-temporal data with geom_glyph_ribbon() and geom_glyph_segment().

These functions create a ribbon geometry designed to display glyphs based on the combination of x_major and y_major. For each x_minor value, geom_glyph_ribbon() displays a y interval defined by ymin_minor and ymax_minor. Meanwhile, geom_glyph_segment() draw a straight line between y_minor and yend_minor with respect to x_minor.

Let’s compare geom_glyph_ribbon() and geom_glyph_segment():

vic_temp <- aus_temp |>
  filter(id %in% c("ASN00026021", "ASN00085291", "ASN00084143"))

# Define a color palette
color_palette <- c("deepskyblue4", "coral3")

p1 <- vic_temp |>
   ggplot(aes(x_major = long,
              y_major = lat,
              x_minor = month,
              ymin_minor = tmin,
              ymax_minor = tmax)) +
  geom_sf(data = abs_ste |> filter(NAME == "Victoria"),
          fill = "antiquewhite", color = "white", inherit.aes = FALSE)  +
  # Customize the size of each glyph box using the width and height parameters.
  add_glyph_boxes(width = rel(2.5), height = rel(1.5),
                  color = color_palette[1]) +
  add_ref_lines(width = rel(2.5), height = rel(1.5),
                color = color_palette[1]) +
  geom_glyph_ribbon(width = rel(2.5), height = rel(1.5),
                    color = color_palette[1], fill = color_palette[1]) +
  # Theme and aesthetic
  theme_glyph() +
  labs(title = "geom_glyph_ribbon()") +
  theme(plot.title = element_text(hjust = 0.5),
        title = element_text(color = color_palette[1],
                             family  = "mono")) 

p2 <- vic_temp |>
   ggplot(aes(x_major = long,
              y_major = lat,
              x_minor = month,
              y_minor = tmin,
              yend_minor = tmax)) +
  geom_sf(data = abs_ste |> filter(NAME == "Victoria"),
         fill = "antiquewhite", color = "white", inherit.aes = FALSE)  +
  # Customize the size of each glyph box using the width and height parameters.
  add_glyph_boxes(width = rel(2.5), height = rel(1.5),
                  color = color_palette[2]) +
  add_ref_lines(width = rel(2.5), height = rel(1.5),
                color = color_palette[2]) +
  geom_glyph_segment(width = rel(2.5), height = rel(1.5),
                    color = color_palette[2]) +
  # Theme and aesthetic
  theme_glyph() +
  labs(title = "geom_glyph_segment()") +
  theme(plot.title = element_text(hjust = 0.5),
        title = element_text(color = color_palette[2]))

grid.arrange(p1, p2, ncol = 2)

Options

Options allow you to modify the behavior of sugarglider to fit the specific needs of your figure. These are all global options that affect every glyph.

Option Default Description
x_scale "identity" This function scales each set of minor values within a grid cell along the x-dimension.
y_scale "identity" This function scales each set of minor values within a grid cell along the y-dimension.
width ggplot2::rel(4) Width of the glyph.
height ggplot2::rel(2.5) Height of the glyph.
global_rescale TRUE Determines whether rescaling is applied globally across all glyphs or individually for each glyph

Aesthetics

sugarglider provides the same aesthetics for geom_glyph_ribbon() and geom_glyph_segment() as those available in geom_ribbon() or geom_segment(), while also introducing additional unique options.

Option Default Description
colour "black" Color for line segments and borders.
linewidth 0.5 Width of the line for borders.
linetype 1 Style of the line for borders.
fill "black" Color of the interior area of the geometries.
alpha 0.8 Transparency level of the glyphs.

Examples

Monthly temperature across Australia

The National Oceanic and Atmospheric Administration (NOAA) provides comprehensive weather data from numerous stations across Australia. The aus_temp dataset includes key climate variables, such as precipitation and temperature, recorded at 29 different weather stations throughout 2020.

head(aus_temp) |>
  kable() |> kable_styling()
id long lat month tmin tmax prcp
ASN00001020 126.3867 -14.09 1 253.4516 319.0000 163.87097
ASN00001020 126.3867 -14.09 2 248.6786 322.6071 162.74074
ASN00001020 126.3867 -14.09 3 253.6129 333.1935 42.00000
ASN00001020 126.3867 -14.09 4 244.0357 340.9310 21.57143
ASN00001020 126.3867 -14.09 5 220.4138 331.9333 0.00000
ASN00001020 126.3867 -14.09 6 202.3667 310.9000 11.20000

Using the default rescaling parameters, we can visualize the temperature data through geom_glyph_segment(), alongside geom_point() elements that mark the location of each weather station. Each segment glyph represents local climate data, offering an intuitive way to explore temperature variations across Australia.

The default identity scaling function is applied to each set of minor values within a grid cell. This method centers the glyphs both vertically and horizontally based on the station’s coordinates and adjusts the minor axes to fit within the interval [-1, 1]. This ensures that the glyphs are appropriately sized to fit the desired dimensions.

aus_temp |>
  ggplot(aes(
    x_major = long, 
    y_major = lat, 
    x_minor = month, 
    y_minor = tmin, 
    yend_minor = tmax)) +
  geom_sf(data = abs_ste, fill = "antiquewhite",
          inherit.aes = FALSE, color = "white") +
  coord_sf(xlim = c(110,155)) +
  # Add glyph box to each glyph
  add_glyph_boxes(
    width = 4,
    height = 3) +
  # Add points for weather station 
  geom_point(aes(x = long, y = lat,
                 color = "Weather Station")) +
  # Customize the size of each glyph box using the width and height parameters.
  geom_glyph_segment(
    width = 4, height = 3,
    aes(color = "Temperature")) +
  # Theme and aesthetic 
  scale_color_manual(
    values = c("Weather Station" = "firebrick",
               "Temperature" = "black")) +
  labs(color = "Data",
       title = "Daily Temperature Variations Across Australian Weather Stations")  +
  theme_glyph()

So far, all the visualizations have used global rescaling (enabled by default), meaning the glyphs are sized relative to one another based on their data values. By disabling global rescaling, we can see the effects of local rescaling, where each glyph is resized based on its individual values.

  • Local Rescale (global_rescale = FALSE): Each line segment’s length is determined by the local temperature range within a region, emphasizing regional differences in temperature patterns.
  • Global Rescale (global_rescale = TRUE): Global temperature range determined the length of each line segment, ensuring that data range remain consistent across all region for easy comparison.

Below is a comparison between the local and global rescaling approaches:

# Global rescale
p1 <- aus_temp |>
  ggplot(aes(
    x_major = long, 
    y_major = lat, 
    x_minor = month, 
    y_minor = tmin, 
    yend_minor = tmax)) +
  geom_sf(data = abs_ste, fill = "antiquewhite",
          inherit.aes = FALSE, color = "white") +
  coord_sf(xlim = c(110,155)) +
  # Add glyph box to each glyph
  add_glyph_boxes() +
  # Add reference lines to each glyph
  add_ref_lines() +
  # Glyph segment plot with global rescale
  geom_glyph_segment(global_rescale = TRUE) +
  labs(title = "Global Rescale") +
  theme_glyph()
  
# Local Rescale
p2 <- aus_temp |>
  ggplot(aes(
    x_major = long, 
    y_major = lat, 
    x_minor = month, 
    y_minor = tmin, 
    yend_minor = tmax)) +
  geom_sf(data = abs_ste, fill = "antiquewhite",
          inherit.aes = FALSE, color = "white") +
  coord_sf(xlim = c(110,155)) +
  # Add glyph box to each glyph
  add_glyph_boxes() +
  # Add reference lines to each glyph
  add_ref_lines() +
  # Glyph segment plot with local rescale
  geom_glyph_segment(global_rescale = FALSE) +
  labs(title = "Local Rescale") +
  theme_glyph()

grid.arrange(p1, p2, ncol = 2)

Highlighting Temperature Changes with Color-Coded Glyph

Expanding on our temperature analysis, we now incorporate precipitation data across Australia using geom_glyph_ribbon(). The glyphs are color-coded to represent varying levels of rainfall, with reference lines and glyph boxes enhancing clarity and allow for easy comparison of precipitation level across the country.

prcp <- aus_temp |>
   group_by(id) |>
   mutate(prcp = mean(prcp, na.rm = TRUE)) |>
   ggplot(aes(x_major = long, y_major = lat,
              x_minor = month, ymin_minor = tmin,
              ymax_minor = tmax, 
              fill = prcp, color = prcp)) +
  geom_sf(data = abs_ste, fill = "antiquewhite",
          inherit.aes = FALSE, color = "white") +
  # Add glyph box to each glyph
   add_glyph_boxes() +
  # Add ref line to each glyph
   add_ref_lines() +
  # Add glyph ribbon plots
   geom_glyph_ribbon() +
   coord_sf(xlim = c(112,155)) +
  # Theme and aesthetic 
  theme_glyph() +
  scale_fill_gradientn(colors = c("#ADD8E6", "#2b5e82", "dodgerblue4")) +
  scale_color_gradientn(colors = c( "#ADD8E6", "#2b5e82", "dodgerblue4")) +
  labs(fill = "Percepitation", color = "Percepitation",
       title = "Precipitation and Temperature Ranges Across Australia") 

prcp

If you’re interested in comparing temperature trends across different years for specific regions in Victoria, geom_glyph_ribbon() provides a way to visualize how temperatures have evolved over time, with each year distinguished by a different color for clarity.


fact <- historical_temp |> 
  filter(id %in% c("ASN00026021", "ASN00085291", "ASN00084143")) |>
   ggplot(aes(color = factor(year), fill = factor(year),
              group = interaction(year,id),
              x_major = long, y_major = lat,
              x_minor = month, ymin_minor = tmin, 
              ymax_minor = tmax)) +
  geom_sf(data = abs_ste |> filter(NAME == "Victoria"),
           fill = "antiquewhite", color = "white",
          inherit.aes = FALSE)  +
  # Customized the dimension of each glyph with `width` and `height` parameters
   add_glyph_boxes(width = rel(2),
                   height = rel(1.5)) +
   add_ref_lines(width = rel(2),
                 height = rel(1.5)) +
   geom_glyph_ribbon(alpha = 0.5,
                     width = rel(2),
                     height = rel(1.5)) +
  labs(x = "Longitude", y = "Latitude",
       color = "year", fill = "year",
       title = "Temperature Trends in Selected Victorian Weather Stations") +
  # Theme and aesthetic
  theme_glyph() +
  theme(legend.position.inside = c(.4,0)) +
  scale_colour_wsj("colors6") +
  scale_fill_wsj("colors6") 

fact

Integrating Glyph Legends

To further enhance map readability, the add_geom_legend() function integrates a larger version of one of the glyphs into the bottom left corner of the plot. This legend helps users interpret the scale of the data.

In the example below, a series of glyph are created using geom_glyph_ribbon() and overlaid on a basemap to depict daily temperature variations across Australian weather stations. A legend is added through add_glyph_legend(), allowing users to easily interpret the range of daily temperature value based on a randomly selected weather station. Since the legend data is drawn from a single, randomly chosen station, it’s important for users to set a seed for reproducibility to ensure consistent results.

set.seed(28493)
legend <- aus_temp |>
   ggplot(aes(x_major = long, y_major = lat,
              x_minor = month, ymin_minor = tmin,
              ymax_minor = tmax)) +
  geom_sf(data = abs_ste, fill = "antiquewhite",
          inherit.aes = FALSE, color = "white") +
  add_glyph_boxes(color = "#227B94") +
  add_ref_lines(color = "#227B94") +
  add_glyph_legend(color = "#227B94", fill = "#227B94") +
  # Add a ribbon legend
  geom_glyph_ribbon(color = "#227B94", fill = "#227B94") +
  theme_glyph() +
  labs(title = "Temperature Ranges Across Australia with Glyph Legend")

legend

Observations and Insights

Both the Geom Glyph Segment and Geom Glyph Ribbon provide valuable insights into seasonal temperature trends across Australia. Disabling global rescaling reveals that most weather stations follow similar curvature trends relative to their neighboring stations. However, with global rescaling enabled, it becomes apparent that coastal regions exhibit far less temperature variation overall.

Flight Variability at U.S. Airports with the Most Cancellations

The U.S. Department of Transportation (DOT) Bureau of Transportation Statistics monitors the on-time performance of domestic flights operated by major U.S. airlines. Each month, the DOT publishes the Air Travel Consumer Report, which provides a summary of on-time, delayed, canceled, and diverted flights.

This dataset, sourced from the Kaggle Airline Flight Delay and Cancellation data, has been processed and aggregated to show the minimum and maximum number of flights originating from the top 10 U.S. airports with the highest cancellation rates.

head(flights) |>
  kable() |> kable_styling()
origin month long lat min_flights max_flights
ATL 1 -84.42806 33.63667 2321 3279
ATL 2 -84.42806 33.63667 1953 3074
ATL 3 -84.42806 33.63667 2667 3476
ATL 4 -84.42806 33.63667 896 3361
ATL 5 -84.42806 33.63667 785 3441
ATL 6 -84.42806 33.63667 1119 3473

In this section of the analysis, we will showcase the functionality of geom_glyph_segment() the monthly range of flights for each airport, providing insights into how flight numbers fluctuate over time. Additionally, we will use geom_glyph_ribbon() to visualize the variation between the minimum and maximum number of flights from each airport.

USmap <- us_map(regions = "state") |>
  filter(full != "Alaska")

# Specify tooltip for ggiraph 
flights <- flights |>
  mutate(tooltip = paste("origin: ",origin,
                         "\nmonth: ", month,
                         "\nmin_flights: ", min_flights,
                         "\nmax_flights: ", max_flights))

fl <- flights |> 
  ggplot(aes(x_major = long, y_major = lat,
             x_minor = month, y_minor = min_flights,
             yend_minor = max_flights,
             tooltip = tooltip)) + 
  geom_sf(data = USmap, color = "white",
          fill = "antiquewhite", inherit.aes = FALSE) +
  coord_sf(crs = st_crs(4326)) +
  add_glyph_boxes(color = "#CD5C08") +
  add_ref_lines(color = "#CD5C08") +
  geom_glyph_segment(color = "#CD5C08") +
  labs(title = "Monthly Flight Variability",
       subtitle = "Based on top 10 US Airports with high cancellations rate") +
  theme_glyph() 

# Interactive plot using ggiraph
girafe(ggobj = fl)

This plot displays the minimum and maximum number of flights departing from each airport on a spatial map. Each line segment represents the flight range for a respective airport. The length of each segment indicates the variability in flight numbers across different airports. Notably, airports like ATL and ORD generally have a higher volume of departing flights compared to others such as MCO and PHX. Additionally, all airports exhibit fluctuations throughout the year, with broader flight intervals mid-year and narrower intervals during the holiday season at year-end.

From the graph, it is evident that airports in certain regions experience more variability than others. While segment plots provide a good estimation of this variability, the geom_glyph_ribbon enhances our understanding by displaying the gap between minimum and maximum flights. Wider ribbons suggest greater fluctuation in flight operations month-to-month.

south <- us_map(include = .south_region)
west <- us_map(include = .west_region, exclude = c("AK", "HI"))
  
  
southR <- flights |> 
  filter(origin %in% c("ATL", "CLT", "MCO", "DFW")) |>
  ggplot(aes(x_major = long, y_major = lat,
             x_minor = month, ymin_minor = min_flights,
             ymax_minor = max_flights)) + 
  geom_sf(data = south, color = "white",
          fill = "antiquewhite", inherit.aes = FALSE) +
  coord_sf(crs = st_crs(4326)) +
  add_glyph_boxes(color = "#A6B37D") +
  add_ref_lines(color = "#A6B37D") +
  geom_glyph_ribbon(color = "#A6B37D", fill = "#A6B37D")  +
  labs(title = "South Region") +
  theme_glyph() 

westR <- flights |> 
  filter(origin %in% c("PHX", "LAS", "LAX", "SEA")) |>
  ggplot(aes(x_major = long, y_major = lat,
             x_minor = month, ymin_minor = min_flights,
             ymax_minor = max_flights)) + 
  geom_sf(data = west, color = "white",
          fill = "antiquewhite", inherit.aes = FALSE) +
  coord_sf(crs = st_crs(4326)) +
  add_glyph_boxes(color = "#CD5C08") +
  add_ref_lines(color = "#CD5C08") +
  geom_glyph_ribbon(color = "#CD5C08", fill = "#CD5C08")  +
  labs(title = "West Region") +
  theme_glyph() 

grid.arrange(westR, southR, ncol = 2)

geom_glyph_ribbon() highlights the disparities in flight volume between regions. The Western region exhibits a larger variation in the number of flights compared to the Southern region, as evidenced by the thicker ribbons. Interestingly, both regions show a notable increase in discrepancies during the mid-year, which supports our findings from the geom_glyph_segment().