24  Animations - Two ways

Today we focus on animating visualizations in R

24.1 Animations

Visualizations that include movement are a another way of creating salience. However, a bad animation doesn’t add anything to the visualization and just requires more time to view the same information than a good visualizationn.

First, let’s show a couple of examples using gganimate which works to extend the grammar of ggplot2. Then we will also show a couple of examples using shiny and its sliderInput() animation.

24.2 Packages and libraries

Install gganimate and gifski. Apple computers may default to image-magick, but I can’t test that.

install.packages('gganimate')
install.packages('gifski') # Note that gifski may require other dependencies such as a Rust installation.

Load libraries used for visualization today.

library(tidyverse) ## or dplyr and tidyr
── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
✔ dplyr     1.1.4     ✔ readr     2.1.5
✔ forcats   1.0.0     ✔ stringr   1.5.1
✔ ggplot2   3.5.1     ✔ tibble    3.2.1
✔ lubridate 1.9.3     ✔ tidyr     1.3.1
✔ purrr     1.0.2     
── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
✖ dplyr::filter() masks stats::filter()
✖ dplyr::lag()    masks stats::lag()
ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errors
Linking to GEOS 3.12.1, GDAL 3.8.4, PROJ 9.3.1; sf_use_s2() is TRUE

24.3 gganimate

24.3.1 Contrived example - Keeling Curve at Mauna Loa

In Lecture 1 we plotted the growing concentration of CO2 at Mauna Loa, the famous Keeling Curve.

Let’s revisit that.

First, import the data from NOAA CMDL.

#read raw data
co2 <- read_table('https://www.gml.noaa.gov/webdata/ccgg/trends/co2/co2_mm_mlo.txt',
                  skip = 57 ) 
#fix column headers
fieldNames <- c('year', 'month', 'decDate', 'meanCO2', 'trendedCO2', 'days', 'stdev', 'unc')
colnames(co2) <- fieldNames
# check dataset back rows
tail(co2)
# A tibble: 6 × 8
   year month decDate meanCO2 trendedCO2  days stdev   unc
  <dbl> <dbl>   <dbl>   <dbl>      <dbl> <dbl> <dbl> <dbl>
1  2024     5   2024.    427.       424.    29  0.76  0.27
2  2024     6   2024.    427.       424.    20  0.65  0.28
3  2024     7   2025.    426.       425.    24  0.69  0.27
4  2024     8   2025.    423.       425.    22  1.08  0.44
5  2024     9   2025.    422.       425.    18  0.41  0.18
6  2024    10   2025.    422.       426.    22  0.35  0.14

Next, let’s create a relatively simple visualization of the Keeling Curve using ggplot2. Figure 24.1 shows the result.

  ggplot(data = co2, aes(x = decDate, y = meanCO2)) +
  geom_point(color = 'black', size = 1) +
  theme_bw() +
  labs(x = 'Year', y = 'Concentration of CO2 (ppm)', 'Keeling Curve @ Mauna Loa')
Figure 24.1: Concentration (ppm) of CO2 measured at Mauna Loa

24.3.2 Add in the animation steps

In many environmental data sets, we will want to show changes over time. gganimate has a built-in function for time animations called [transition_time()](https://gganimate.com/reference/transition_time.html).

Let’s do the most super-basic animation and add that function to the Keeling Curve visualization. Figure 24.2 shows the most basic animation when adding a time increment.

ggplot(data = co2, aes(x = decDate, y = meanCO2)) +
  geom_point(color = 'black', size = 1) +
  theme_bw() +
  labs(x = 'Year', y = 'Concentration of CO2 (ppm)', 'Keeling Curve @ Mauna Loa') +
  transition_time(year) # This step adds the by year animation.
Figure 24.2: Animated Concentration (ppm) of CO2 measured at Mauna Loa

Pretty cool, but we don’t see the old data so it looks just like a migrating flock of points. If we want to show other points along the graph, we can use shadow_mark() to show other points along the graph. Arguments for past and future allow us to choose include either or both of those points.

shadow_mark(past = TRUE, future = FALSE, ..., exclude_layer = NULL)

Figure 24.3 shows the result, while adding in a color argument to shadow_mark to show the old data differently.

ggplot(data = co2, aes(x = decDate, y = meanCO2)) +
  geom_point(color = 'black', size = 1) +
  theme_bw() +
  labs(x = 'Year', y = 'Concentration of CO2 (ppm)', 
       title = 'Year: {frame_time}') +
  transition_time(year) +
  shadow_mark(past = TRUE, color = 'gray') # this leaves behind the animated data already shown.  Feel free to choose a different color.
Figure 24.3: Animated Concentration (ppm) of CO2 measured at Mauna Loa with shadowed past

Pretty close, but that Year title is horrible and the ten significant figures past the decimal point is very aesthetically displeasing. I must fix that using the round() function.

The interesting thing about that curly bracket notation is it can deal with variables and code directly. So let’s modify that directly.

Figure 24.4 shows the fixed title.

ggplot(data = co2, aes(x = decDate, y = meanCO2)) +
  geom_point(color = 'black', size = 1) +
  theme_bw() +
  labs(x = 'Year', y = 'Concentration of CO2 (ppm)', 
       title = 'Year: {round(frame_time, 0)}') + # the rounded value of the year is inserted here, with zero to make it an integer.
  transition_time(year) +
  shadow_mark(past = TRUE, color = 'grey')
Figure 24.4: Animated Concentration (ppm) of CO2 measured at Mauna Loa with shadowed past and rounded Year in title.

24.3.3 Example 2: Animating a ggplot map

Import warehouse data for Riverside County only - let’s limit the scope.

WH.url <- 'https://raw.githubusercontent.com/RadicalResearchLLC/WarehouseMap/main/WarehouseCITY/geoJSON/comboFinal.geojson'
warehouses <- st_read(WH.url) |>  
  filter(county == 'Riverside County') |>  
  st_transform(crs = 4326)
Reading layer `comboFinal' from data source 
  `https://raw.githubusercontent.com/RadicalResearchLLC/WarehouseMap/main/WarehouseCITY/geoJSON/comboFinal.geojson' 
  using driver `GeoJSON'
Simple feature collection with 9084 features and 8 fields
Geometry type: MULTIPOLYGON
Dimension:     XY
Bounding box:  xmin: -118.8037 ymin: 33.43325 xmax: -114.4085 ymax: 35.55527
Geodetic CRS:  WGS 84
head(warehouses)
Simple feature collection with 6 features and 8 fields
Geometry type: MULTIPOLYGON
Dimension:     XY
Bounding box:  xmin: -117.6015 ymin: 33.8773 xmax: -117.5314 ymax: 33.97265
Geodetic CRS:  WGS 84
        apn shape_area category year_built                 class
1 115050036     343900 Existing       2000 warehouse/dry storage
2 115060057     206900 Existing       1980 warehouse/dry storage
3 115670012      71800 Existing       1999 warehouse/dry storage
4 144010064      93200 Existing       2018 warehouse/dry storage
5 144010070     206100 Existing       2018 warehouse/dry storage
6 144010076     190000 Existing       1980 warehouse/dry storage
            county unknown place_name                       geometry
1 Riverside County   FALSE     Corona MULTIPOLYGON (((-117.5428 3...
2 Riverside County    TRUE     Corona MULTIPOLYGON (((-117.5533 3...
3 Riverside County   FALSE     Corona MULTIPOLYGON (((-117.5314 3...
4 Riverside County   FALSE   Eastvale MULTIPOLYGON (((-117.5946 3...
5 Riverside County   FALSE   Eastvale MULTIPOLYGON (((-117.6003 3...
6 Riverside County    TRUE   Eastvale MULTIPOLYGON (((-117.5962 3...

Make a basic warehouse map near my house using ggplot and geom_sf. Figure 24.5 shows a basic map of warehouses in ggplot.

ggplot(data = warehouses) +
  geom_sf() +
  coord_sf(xlim = c(-117.35, -117.1),
           ylim = c(33.8,33.95), crs = 4326) +
  theme_void()
Figure 24.5: A simple map of warehouses near Mike’s house

Let’s animate it. We’ll add a second step to control the animation speed and frames. First, we add transition_time() and shadow_mark() in a way identical to our CO2 figure.

Pass the ggplot code chunk into a variable. This variable is then run through an animate() function to control the frame rate and number of frames displayed.

Figure 24.6 shows the time series animation.

data4map <- ggplot(data = warehouses) +
  geom_sf(fill = 'black') +
  coord_sf(xlim = c(-117.35, -117.1),
           ylim = c(33.8,33.95), crs = 4326) +
  theme_void() + 
  transition_time(year_built) + # new line to animate by year_built
  shadow_mark(past = TRUE, color = 'grey20', fill = 'grey') + # new line to leave built warehouses as grey blocks
  labs(title = 'Year: {round(frame_time, 0)}') # new title for figure showing the rounded year year

animate(data4map, nframes = 46, fps = 3, end_pause = 10) # this is a wrapper to control the speed of the animation.
Figure 24.6: An animated map of warehouses growing near Mike’s house

Excellent! We can also add an underlying map of jurisdictions or a tile layer to make it a bit prettier.

Let’s use tigris places function to add some jurisdictions.

library(tigris)

#read the data
Riverside_places <- places(state = 'CA', cb = T, #pull for California, lower detail level
                    year = 2022, 
                    progress_bar = FALSE) |> 
  filter(NAME %in% c('Moreno Valley', 'Mead Valley', 'March ARB', 'Perris', 'Riverside', 'Nuevo')) |> 
  st_transform(crs = 4326)
  
head(Riverside_places)
Simple feature collection with 6 features and 12 fields
Geometry type: MULTIPOLYGON
Dimension:     XY
Bounding box:  xmin: -117.5239 ymin: 33.71387 xmax: -117.0883 ymax: 34.01935
Geodetic CRS:  WGS 84
  STATEFP PLACEFP  PLACENS         AFFGEOID   GEOID          NAME
1      06   62000 02410965 1600000US0662000 0662000     Riverside
2      06   52624 02408960 1600000US0652624 0652624         Nuevo
3      06   45680 02408176 1600000US0645680 0645680     March ARB
4      06   46646 02583077 1600000US0646646 0646646   Mead Valley
5      06   56700 02411403 1600000US0656700 0656700        Perris
6      06   49270 02411159 1600000US0649270 0649270 Moreno Valley
            NAMELSAD STUSPS STATE_NAME LSAD     ALAND AWATER
1     Riverside city     CA California   25 210174364 797658
2          Nuevo CDP     CA California   57  17531432      0
3      March ARB CDP     CA California   57  30946983  36554
4    Mead Valley CDP     CA California   57  49432252      0
5        Perris city     CA California   25  81748619 279334
6 Moreno Valley city     CA California   25 132891703 546511
                        geometry
1 MULTIPOLYGON (((-117.5238 3...
2 MULTIPOLYGON (((-117.1695 3...
3 MULTIPOLYGON (((-117.3194 3...
4 MULTIPOLYGON (((-117.3313 3...
5 MULTIPOLYGON (((-117.2627 3...
6 MULTIPOLYGON (((-117.2965 3...

And this chapter from the ggplot2 book is much better than my lecture on color - please review.

data4map2 <- ggplot() +
  geom_sf(data = Riverside_places, aes(fill = NAME), alpha = 0.2) +
  geom_sf(data = warehouses, color = 'black', fill = '#653503',
          inherit.aes = FALSE) +
  coord_sf(xlim = c(-117.35, -117.1),
           ylim = c(33.7,33.95), crs = 4326) +
  theme_void() + 
  transition_time(year_built) +
  shadow_mark(past = TRUE, color = '#653503', fill = 'grey') +
  labs(title = 'Year: {round(frame_time, 0)}') +
  scale_fill_brewer(palette = 'Accent')

animate(data4map2, nframes = 47, fps = 3, end_pause = 10)
Figure 24.7: An animated map of warehouses growing near Mike’s house with a background layer

It could still be better adding roads and rail and other features and making the colors less ridiculous, but this is a good enough example for today.

24.4 Animations in Shiny using sliderInput()

24.4.1 Example 3 - Animate Old Faithful Histogram

Create a new shiny App called ‘animate.R’ as shown in Section 19.3.2.

This should create a new Shiny App using the Old Faithful Geyser data. We’re going to animate this using the following argument within the sliderInput() function.

animate = TRUE.

The code chunk within the app should look like this, starting at line 21 on my machine:

            sliderInput("bins",
                        "Number of bins:",
                        min = 1,
                        max = 50,
                        value = 30,
                        # NEW ARGUMENT HERE!
                        animate = TRUE)

If we run the app by pressing the Run App button, a shiny App should pop-up. I’ll show that within the Shiny App. A blue play button will appear on the bottom-right of the slider. Pressing the play button advances through the slider increments and updates the histogram.

Easy!

There are additional options for controlling the interval rate and whether it loops.

We can modify the code to show that.

            sliderInput("bins",
                        "Number of bins:",
                        min = 1,
                        max = 50,
                        value = 30,
                        # NEW ARGUMENT HERE!
                        animate = animationOptions(
                          loop = TRUE, interval = 300)
                        )