Graph info

Should I use this graph?


This graph requires:

✅ a categorical variable (preferably a factor)

✅ a numeric (continuous) variable

✅ a numeric (date) variable

Description

Stream graphs display how a numerical variable (typically on the y axis) changes over time (on the x axis) across levels of a categorical variable. These graphs are handy if the numerical value varies wildly (or isn’t always present) over the time measurement.

Categorical groups are differentiated by color layers, with the area of the layer representing the change in y value. In ggplot2, we can create stream graphs using ggstream.

Getting set up

PACKAGES:

Install packages.

Code 
remotes::install_github("davidsjoberg/ggstream")
install.packages("ggplot2movies")
library(ggstream)
library(ggplot2movies)
library(ggplot2)

DATA:

We’re going to use only the mpaa, year, and budget columns from ggplot2movies::movies, then drop all missing values (we have to remove special missing characters from mpaa).

We’ll then convert mpaa to an ordered factor, then group by year and mpaa to calculate the average budget and filter to only those movies after 1984.

Code 
movies_stream <- ggplot2movies::movies |> 
    dplyr::select(mpaa, year, budget) |> 
    tidyr::drop_na() |> 
    dplyr::filter(mpaa != "") |> 
    dplyr::mutate(mpaa = factor(mpaa, 
                            levels = c("NC-17", "R", 
                                       "PG-13", "PG"), 
                            ordered = TRUE)) |> 
    dplyr::group_by(year, mpaa) |> 
    dplyr::summarise(
        avg_budget = mean(budget, na.rm = TRUE)) |> 
    dplyr::ungroup() |> 
    dplyr::filter(year > 1984)
dplyr::glimpse(movies_stream)
Rows: 47
Columns: 3
$ year       <int> 1986, 1989, 1989, 1990, 1991, 1991, 1992, 1993, 1993, 1994,…
$ mpaa       <ord> R, R, PG-13, R, R, PG, R, R, PG-13, R, PG-13, PG, NC-17, R,…
$ avg_budget <dbl> 17250000, 787000, 39250000, 3866667, 35750000, 300000, 1475…

The grammar

CODE:

Create labels with labs()

  • use paste0() in the subtitle to automatically update the year when the movies_stream changes

Initialize the graph with ggplot() and provide data

Map year to the x, avg_budget to y, and mpaa to fill

Add the geom_stream() layer

Adjust the y axis with scale_y_continuous() and scales::dollar

Finally, move the legend with theme(legend.position = "bottom")

Code 
labs_stream <- labs(
    title = "20 years of movie budgets", 
    subtitle = 
        paste0("movies between", 
            min(movies_stream$year), 
            " and ", 
            max(movies_stream$year)),
    x = "Year", 
    y = "Average Movie Budget")

ggp2_stream <- ggplot(data = movies_stream, 
    mapping = aes(x = year, 
                  y = avg_budget, 
                  fill = mpaa)) + 
    ggstream::geom_stream() + 
    scale_y_continuous(labels = scales::dollar)
ggp2_stream + 
    labs_stream + 
    theme(legend.position = "bottom")

GRAPH:

More info

The ggstream package has multiple arguments for adjusting the shape and look of the categorical levels (and text labels).

TYPE:

Map mpaa to fill (wrapped in forcats::fct_rev())

We can adjust the look of the graph by setting the type argument in geom_stream()

  • type: change type from "mirror" (the default) to "proportional"

Adjust colors:

  • scale_fill_manual(): Add colors as a named vector to values

Finally, add the fill to the labels and move the legend with theme(legend.position = "bottom")

Code 
ggp2_stream_prp <- ggplot(data = movies_stream, 
    mapping = aes(x = year, 
                  y = avg_budget, 
                  fill = forcats::fct_rev(mpaa))) + 
    ggstream::geom_stream(type = "proportional") + 
    scale_y_continuous(labels = scales::percent) + 
  scale_fill_manual(
      values = c("PG-13" = "#0bd3d3", 
                 "PG" = "#f890e7", 
                 "R" = "#fa7b3c", 
                 "NC-17" = "#772953"))
ggp2_stream_prp + 
    labs_stream + 
    labs(fill = "mpaa") + 
    theme(legend.position = "bottom")

SORTING:

To change how the categorical areas are sorted, adjust the sorting argument to either "none", "onset", or "inside_out"

  • sorting: set the sorting method to "inside_out" in both ggstream::geom_stream() and ggstream::geom_stream_label()
  • type: change the type to "ridge" in both ggstream::geom_stream() and ggstream::geom_stream_label()

We can also add text labels using ggstream::geom_stream_label():

  • Map mpaa (wrapped in forcats::fct_rev()) to label globally
  • Inside ggstream::geom_stream_label():
    • set color to white ("#ffffff") and the size to 2.7
    • Remove the legend with show.legend = FALSE

Colors:

  • Use scale_colour_manual() and scale_fill_manual() to manually set the values using a named vector

Change the y axis to US dollars using scale_y_continuous() and scales::dollar

Finally, add the fill to the labels and move the legend with theme(legend.position = "bottom")

Code 
ggp2_stream_srt <- ggplot(data = movies_stream, 
    mapping = aes(x = year, 
                  y = avg_budget, 
                  fill = fct_rev(mpaa),
                  label = fct_rev(mpaa))) + 
    ggstream::geom_stream(
        type = "ridge", 
        sorting = "inside_out") + 
    ggstream::geom_stream_label(
        type = "ridge", 
        sorting = "inside_out", 
        color = "#ffffff", 
        size = 2.7,
        show.legend = FALSE) +
    scale_colour_manual(
      values = c("PG-13" = "#0bd3d3", 
                 "PG" = "#f890e7", 
                 "R" = "#fa7b3c", 
                 "NC-17" = "#772953")) +
    scale_fill_manual(
      values = c("PG-13" = "#0bd3d3", 
                 "PG" = "#f890e7", 
                 "R" = "#fa7b3c", 
                 "NC-17" = "#772953")) + 
    scale_y_continuous(labels = scales::dollar) 
    
ggp2_stream_srt + 
    labs_stream + 
    labs(fill = "mpaa") + 
    theme(legend.position = "bottom")