Explore spatial patterns of summer sea surface temperature in the Baltic Sea graphically:

How to get from left to right?

p <- ggplot(hydro, aes(long, lat))
str(p)

ggplot(data, aes(x, y, color, size, shape, linetype, fill, alpha, label, family, width, hjust, vjust, ...))

• You can display a point in different ways by changing the values of its aesthetics (e.g. size, shape, and colour of border and inside):

• Use aes() to associate the name of the aesthetic with a variable to display, for single values aes() not needed
• Once you map an aesthetic, ggplot2 takes care of the rest:
  • selects a reasonable scale
  • constructs a legend that explains the mapping between levels and values
• For x and y aesthetics, ggplot2 does not create a legend, but it creates an axis line with tick marks and a label

For some geoms/point shapes you can specify 2 colour aesthetics: the border colour (=colour) and the inside (=fill)

ggplot(hydro, aes(x = fmonth)) +
  geom_bar(aes(colour = fmonth), fill = "white")

To make colours transparent use alpha aesthethics

• range from 0 = transparent to 1 = opaque

col_df <- data.frame(x = 1:5, y = 1:5)
ggplot(col_df, aes(x, y)) +
  geom_point(fill = "red", 
    alpha = c(.1,.3,.5,.7,1.0), 
    size = 5, shape = 21) 

Default shape is 19

shapes <- data.frame(
  shape = c(0:19, 22, 21, 24, 23, 20),
  x = 0:24 %/% 5, y = -(0:24 %% 5))

ggplot(shapes, aes(x, y)) + 
  geom_point(aes(shape = shape),  
    size = 10, fill = "red") +
  geom_text(aes(label = shape), 
    hjust = 0, nudge_x = 0.15) +
  scale_shape_identity() +
  expand_limits(x = 4.1) +
  scale_x_continuous(NULL,breaks=NULL) + 
  scale_y_continuous(NULL,breaks=NULL)

Line types can be specified with an integer or name: 0 = blank, 1 = solid, 2 = dashed, 3 = dotted, 4 = dotdash, 5 = longdash, 6 = twodash, as shown below:

lty <- c("blank", "solid", "dashed", 
  "dotted", "dotdash", 
  "longdash","twodash")
linetypes <- data.frame(
  y = seq_along(lty), lty = lty) 
ggplot(linetypes, aes(0, y)) + 
  geom_segment(aes(xend = 5, 
    yend = y, linetype = lty)) + 
  scale_linetype_identity() + 
  geom_text(aes(label = lty), 
    hjust = 0, nudge_y = 0.2) +
  scale_x_continuous(NULL,breaks=NULL) + 
  scale_y_continuous(NULL,breaks=NULL)

There are only three fonts that are guaranteed to work everywhere: "sans" (the default), "serif", or "mono":

df <- data.frame(x = 1, y = 3:1, 
  family = c("sans", "serif", "mono"))
ggplot(df, aes(x, y)) + 
  geom_text(aes(label = family, 
    family = family), size = 15)

Horizontal and vertical justification have the same parameterisation, either a string (“top”, “middle”, “bottom”, “left”, “center”, “right”) or a number between 0 and 1:

just <- expand.grid(hjust = c(0,0.5,1), 
  vjust = c(0, 0.5, 1))
just$label <- paste0(just$hjust, ", ", 
  just$vjust)

ggplot(just, aes(hjust, vjust)) +
  geom_point(colour = "grey70", 
    size = 5) + 
  geom_text(aes(label = label, 
    hjust = hjust, vjust = vjust),
    size = 7)

• To change the default settings of an aesthetic use a scale that applies to that mapping (if you have used aes() for a specific aesthetic):
  • aesthetic mapping = What variable to map to e.g. colour
  • scale = How to map the variable to e.g. color
• Use the three part scale name convention and set additional arguments in the scale function if desired:

ggplot2's default discrete palettes (depends on how many colors are needed)

Example of oceanographic stations sampled in summer months

sst_sum <- hydro %>% filter(pres == 1, fmonth %in% 5:8)
sst_sum$fmonth <- factor(sst_sum$fmonth) # to remove other factor levels
p <- ggplot(sst_sum, aes(x = long, y = lat, colour = fmonth)) + geom_point()

scale_colour_hue is the default colour scale → does not generate colour-blind safe palettes

p + scale_colour_hue()

p + scale_colour_hue(l = 50, c = 150, 
  h = c(150, 270) )

Example of summer sea surface temperature (SST)

# Same dataset as before
sst_sum <- hydro %>% filter(pres == 1, fmonth %in% 5:8)
sst_sum$fmonth <- factor(sst_sum$fmonth)
p <- ggplot(sst_sum, aes(x = long, y = lat, colour = temp)) + geom_point()

scale_colour_gradient is the default colour scale. Creates a two colour gradient (low-high):

p

p + scale_colour_gradient()

Change a few settings ...

p + scale_colour_gradient(
  low = "yellow", high = "red")

p + scale_colour_gradient(
  low = "white", high = "maroon4", 
  na.value = "yellow")

Creates a diverging colour gradient (low-mid-high):

p + scale_colour_gradient2(midpoint=10)

p + scale_colour_gradient2(midpoint=15, 
  low = "blue", mid = "yellow", 
  high = "red", na.value = "black")

Example of sea surface temperature (SST) in June

sst_june <- hydro %>% filter(pres == 1, fmonth == "6")

You can use scale_size(), scale_size_area(), or scale_radius():

p <- ggplot(sst_june, aes(long,lat))+ 
  geom_point(aes(size = temp),
    colour = "orange2", alpha=0.3)

p + scale_radius(range=c(1,10), 
    breaks=seq(5,20,2.5), limits=c(4,20))

There are scales for every aesthetic ggplot2 uses. To see a complete list with examples, visit http://docs.ggplot2.org/current

Screenshot from webpage (taken Dec. 2017)

Facets divide a plot into subplots based on the values of one or more discrete variables

To adjust facet labels set labeller

p + facet_grid(pres ~ fmonth, labeller = label_both)

• Determine the coordinate plane to draw the graph on.
• Use the two part scale name convention and set additional arguments in the coord_function if desired:

Which months are most frequently sampled?

In the grammar of graphics, a pie chart is a stacked bar graph in polar coordinates.

monthly_pie <- hydro %>% 
  select(fmonth, station, cruise, 
    date_time) %>% distinct() %>% 
  group_by(fmonth) %>% 
  mutate(year = "y")

ggplot(monthly_pie, aes(year)) + 
  geom_bar(aes(fill = fmonth)) + 
  scale_fill_brewer(
    palette="Paired") +
  coord_polar(theta="y", direction=-1)

coord_map() projects a portion of the earth, which is approximately spherical, onto a flat 2D plane using any projection defined by the mapproj package.

# Showing the world map:
world <- map_data("world")
worldmap <- ggplot(world, aes(
    x=long, y=lat, group=group)) +
  geom_polygon(fill = "white", 
    colour = "black") 

worldmap + coord_map("ortho", 
  orientation = c(55, 20, 0))

Simply add the axis limits to coord_map()

worldmap + coord_map("ortho", 
  orientation = c(55, 20, 0),
  xlim = c(10, 30), ylim = c(54,66))

• Good labels are critical for making your plots understandable to a wider audience.
• Both axis and legend labels should show the full variable name or a commonly used abbreviation.
• The title and subtitle should provide main findings.

• Good labels are critical for making your plots understandable to a wider audience.
• Both axis and legend labels should show the full variable name or a commonly used abbreviation.

• The title and subtitle should provide main findings.

• There are different functions available to change labels:

  • xlab(): change the label of the x-axis
  • ylab(): change the label of the y-axis
  • ggtitle(): provide a title (and subtitle)
  • labs(): change all labels

• Control over the legends is decentralized:
  • Choose the type of legend for each aesthetic with guides()
  • Make adjustments to the legend with guide_legend() or guide_colourbar(), which can be specified in each scale_* function or in guides():

→ Lets get to the last topic ...

• A theme determines how the plot looks.
• ggplot2 comes with several pre-loaded themes that control the appearance of non-data elements.
• Customization → You can modify individual details of the current theme with theme() or even create your own theme.

Change the legend position with theme()

p3 + theme(legend.position = "bottom", 
  legend.justification = "left") 

p3 + theme(
  legend.position = c(0.01, .95),
  legend.justification = c("left", "top"),
  legend.box.just = "left"
)

See for more info

• ?theme and
• Winston Chang's wiki on github

• ggplot has 7 themes in-built, try them out yourself:

p3 + theme_bw()
p3 + theme_gray()
p3 + theme_classic()
p3 + theme_light()
p3 + theme_linedraw()
p3 + theme_minimal()
p3 + theme_void()

ggthemes by Jeffrey Arnold

• https://github.com/jrnold/ggthemes
• ggthemes vignette
• install.packages("ggthemes")

ggtech by Ricardo Bion

• https://github.com/ricardo-bion/ggtech
• devtools::install_github("ricardo-bion/ggtech", dependencies=TRUE)
• ggtech offers themes that resemble google, airbnb, twitter, facebook
  
• You need to install various necessary fonts manually before using ggtech!

• For ggplot objects the standard R functions par() and layout() do not work
• Alternative functions:
  • viewport() and grid.layout() in the grid package
  • grid.arrange() in the gridExtra package
  • plot_grid() and draw_plot() in the cowplot package
• All these function use internally the following grid specifications (in "npc" units):

• For ggplot objects the standard R functions par() and layout() do not work
• Alternative functions:
  • viewport() and grid.layout() in the grid package
  • grid.arrange() in the gridExtra package
  • plot_grid() and draw_plot() in the cowplot package
• All these function use internally the following grid specifications (in "npc" units):
• Each approach will be demonstrated using 3 different plots from the hydro dataset:
  • a: barplot showing the number of samplings per month
  • b: scatterplot showing the temperature ~ salinity relationship
  • c: monthly maps of temperature conditions

Demo plots, each assigned to an object

a <- hydro %>% 
  select(fmonth, station, cruise, date_time) %>%
  distinct()  %>% group_by(fmonth) %>%
  ggplot(aes(x = fmonth)) + 
    geom_bar(aes(fill = fmonth)) + 
    guides(fill = "none") +
    scale_fill_brewer(palette = "Paired")

b <- ggplot(hydro, aes(x = psal, y = temp, col = day)) +
  geom_point()

c <- hydro %>% 
  filter(pres %in% c(1, 50), fmonth %in% 5:8) %>%
  ggplot(aes(long,lat))+ 
    geom_point(aes(colour = temp)) +
    scale_colour_gradient(low = "white", high = "red") +
    facet_grid(. ~ fmonth) 

To have a small subplot embedded drawn on top of the main plot

1. first plot the main plot,
   
2. then draw the subplot in a smaller viewport using also the print()function

The viewport() function, has as arguments x, y, width and height to control the position and size of the viewport.

library(grid)
# specify the viewport for the subplot
subvp <- viewport(x = 0.7, y = 0.8, 
  width = 0.3, height = 0.3)
b
print(a, vp = subvp)

• grid.arrange() set up a so-called gtable layout to place multiple grobs on a page in the current device.
• Takes for the grid layout number of rows and columns as arguments:

library(gridExtra)
grid.arrange(a, b, c, 
  nrow = 2, ncol = 2)

• To change the row or column span of a plot, use the function arrangeGrob() when listing the plotting objects. This function takes also the row or column number as input:

grid.arrange(
  arrangeGrob(a, b, ncol = 2), 
  # a and b will be in the 1st row,  
  # splitted into 2 columns  
  c, # c will be in the 2nd row
  nrow = 2
  )

• It is possible to use the layout_matrix argument for changing the row or column span of a plot (more flexible than arrangeGrob()):

grid.arrange(a, b, c, 
  nrow = 2, ncol = 2,
  layout_matrix = matrix(
    c(1,2,2,3,3,3), 
    nrow = 2, byrow = TRUE))

• provides a publication-ready theme for ggplot2
• has its own built-in default theme: white background and no grid (similar to theme_classic()), different font sizes
• plot_grid() is a shortcut function with limited adjustments

library(cowplot)
plot_grid(a,b,c, labels = c("a)", "b)", "c)"), ncol = 3)

For more adjustments (location and size) use in combination:

1. ggdraw() → initializes an empty drawing canvas
2. draw_plot() → places a single plot onto the canvas
3. draw_plot_label() → adds labels to the plots (default is upper left corner)

abc <- ggdraw() +
  draw_plot(a, x = 0, y = .5, width = .5, height = .5) +
  draw_plot(b, x = .5, y = .5, width = .5, height = .5) +
  draw_plot(c, x = 0, y = 0, width = 1, height = .5) +
  draw_plot_label(label = c("a)", "b)", "c)"),
    x = c(0, 0.5, 0), y = c(1, 1, 0.5), size = 15)
abc

• Recall: ggsave() (in ggplot2) can be used to save ggplots as pdf.
• A better solution for cowplots: save_plot() (in cowplot) → grid layout can be specified so that the output pdf is nicely formatted and scaled

save_plot("abc3.pdf", abc, 
  base_aspect_ratio = 1.3, # this makes room for a figure legend
  # the grid specification adjusts the scales:
  ncol = 2,  nrow = 2 )

Explore spatial patterns of summer sea surface temperature in the Baltic Sea graphically:

How to get from left to right?

Try to reproduce some of the plots in this presentation and the quiz and read chapter 3 on data visualization in 'R for Data Science' or the book ggplot2: Elegant Graphics for Data Analysis by H. Wickham.

Then go grab a coffee, lean back and enjoy the rest of the day...!

world <- map_data("world")
worldmap <- ggplot(world, aes(x = long, y = lat)) +
  geom_polygon(aes(group = group), fill = "ivory3", colour = "black") 
baltic <- worldmap + coord_map("ortho",  xlim = c(10, 30), ylim = c(54,66))

baltic + geom_point(data = sst_sum, aes(x = long, y = lat, colour = temp), size = 1) +
  scale_colour_gradient(low = "white", high = "red") +
  guides(colour = guide_colourbar(title = "SST (in °C)")) +
  facet_wrap(~fmonth, labeller = label_both) +
  ggtitle(label = "Thermal pattern in the Baltic Sea",
    subtitle = "SST in the summer months of 2015") +
    xlab("Longitude") + ylab("Latitude") +
  ggthemes::theme_base() +
  theme(legend.position = "bottom", legend.title.align = 1,
    legend.title = element_text(colour = "red", angle = 0), 
      panel.background = element_rect(fill = "lightblue1")
  )