Review

• A data visualisation consists of data symbols, guides, and labels.

• We produce data symbols by mapping data values to the visual features of a shape.

• Which visual feature we choose depends on:

  • Whether we have quantitative or qualitative data.
  • The accuracy and capacity of the visual feature.
  • The question we are interested in answering.

Text

• We can map data values to text data symbols and text on guides (axes and legends).

• We can map metadata and data statistics to text labels (titles, captions, and annotations).

Text

• The importance of text

• Text data symbols

• Text Features

• Guides

• Labels

The Importance of Text

• Without text, it is difficult to extract any useful information from a data visualisation.

  

The Importance of Text

• Viewers pay a lot of attention to the text on a data visualisation.

The Importance of Text

• Viewers pay a lot of attention to the text on a data visualisation.

The Visual System

• Text is initially processed as visual information, but is then processed as verbal information.

• Text consists of one or more familiar visual objects.

A Simple Model of the Visual System

Youth Crime

• The crimeDistrictTotal data frame contains the total of offenders by district.

  head(crimeDistrictTotal)

  ## # A tibble: 6 × 3
  ##   district         total avgPop
  ##   <chr>            <int>  <dbl>
  ## 1 Auckland City     5583 17438.
  ## 2 Bay of Plenty     9369 15192.
  ## 3 Canterbury        8976 23550.
  ## 4 Central           8700 15491.
  ## 5 Counties/Manukau 10229 26388.
  ## 6 Eastern           6221  9382.

Text as a Data Symbol

• We can map each total value to the shape of a text data symbol.

  

Text as a Data Symbol

• We can map back from text data symbols to the raw data.

  

Text as a Data Symbol

• Text is appropriate for representing quantitative data.

  • Text has excellent accuracy, for single data values.

  

Text as a Data Symbol

• Differences and ratios can be calculated, but that involves much greater cognitive load.

  

Text as a Data Symbol

• Text is appropriate for representing qualitative data.

  • Text has excellent capacity.

  

• Text can express zero and negative values.

• Text has spectacular congruence.

  

Text as a Data Symbol

• Text data symbols are terrible for visual summaries/data statistics.

  

Text as a Data Symbol

• Text data symbols are terrible for visual summaries/data statistics.

  

Text Data Symbols in {ggplot2}

• We can map data values to text with geom_text().

  ggplot(crimeDistrictTotal) +
      geom_text(aes(x="", y=district, label=total), 
                size=4, hjust=1, vjust=0)

  

   

  • size is in millimetres.
  • hjust is horizontal justification
    (0 = left, 1 = right)
  • vjust is vertical justification
    (0 = bottom, 1 = top)

Text Features

• We can also map data values to other visual features of text data symbols: position, size, angle, and colour

  

Text Features

• A text data symbol often involves a redundant mapping.

  

Fonts

• The shape of text is also affected by the font.

  

Text Features in {ggplot2}

• The font family describes an overall style.

• The fontface describes whether the text is bold, italic, or plain.

  ggplot(pets) + 
      geom_col(aes(pet, count)) +
      geom_text(aes(pet, 150, label=pet),
                colour="white", size=c(24, 30), vjust=0,
                family=c("Impact", "TeX Gyre Chorus"), 
                fontface=c("bold", "italic")) 

Text Features in {ggplot2}

• The font families "sans", "serif", and "mono" are always available.

• Selecting a custom font family is easy as long as you are using the right graphics device:

  • Cairo-based devices.
  • {ragg} devices.

• Further reading:

  • Fonts in R
  • Graphics devices in R

Guides

• Guides are visual representations of scales, such as axes and legends.

  

Guides

• Data values are mapped to data symbols such as lines to create tick marks, grid lines, and legends keys

  

Guides

• Data values are mapped to text data symbols to create tick labels and legends labels

  

Guides in {ggplot2}

• Axes and legends are generated automatically.

• The scale functions like scale_x_continuous() and scale_colour_manual() allow control of the details.

  • breaks specifies the tick mark locations.
  • labels specifies the tick labels, either explicitly or with a function like scales::label_comma().
  • guide takes a function like guide_axis() or guide_legend() to control the placement and layout of the axis or legend.

Guides in {ggplot2}

ggplot(crimeDistrict) +
    geom_line(aes(x=year, y=count, colour=district)) +
    scale_y_continuous(breaks=seq(200, 1400, 200),
                       labels=scales::label_comma())

Guides in {ggplot2}

ggplot(crimeDistrict) +
    geom_line(aes(x=year, y=count, colour=district)) +
    scale_colour_discrete(guide=guide_legend(ncol=2))

Direct Labelling

• Direct labelling draws text data symbols in proximity to other data symbols rather than on a separate guide.

  

Direct Labelling in {ggplot2}

ggplot(crimeDistrict) +
    geom_text(data=subset(crimeDistrict, year == 2011),
              aes(x=year - .1, y=count, label=district, colour=district),
              hjust=1) +
    geom_line(aes(x=year, y=count, colour=district))

{ggrepel}

ggplot(crimeDistrict) +
    geom_text_repel(data=subset(crimeDistrict, year == 2011),
                    aes(x=year-.1, y=count, label=district, colour=district),
                    hjust=1, direction="y") +
    geom_line(aes(x=year, y=count, colour=district)) 

{directlabel}

gg <- ggplot(crimeDistrict) +
    geom_line(aes(x=year, y=count, colour=district)) 

direct.label(gg, "first.points")

Labels

• Labels include titles, captions, and annotations

  

Plot Title

• A plot title may describe the overall purpose of the plot and may describe the overall message of the plot.

  

Captions

• A caption may provide a more detailed description of the plot and may provide information about the data source.

  

Guide Titles

• Axis titles and guide titles describe the variables that are being mapped and can include metadata such as the units of measurement.

  

Annotations

• An annotation may be used to describe an important feature within a plot.

  

Labels

• We can map back from labels to metadata.

  

• Labels are excellent for visual summaries/data statistics.

• Labels and data symbols often work to support each other.

  

Labels

• We can map back from labels to data statistics.

  

Labels in {ggplot2}

• The labs() function can be used to specify the plot title and caption.

  labels <- 
      labs(title=paste("Youth Crime has Declined",
                       "in all Districts over the last Decade"),
           subtitle="(with small uptick in the last two years)",
           caption=paste('Number of youth offenders aged between',
                         '14 and 16 from 2010 to 2020 in',
                         'New Zealand.\nSource: New Zealand',
                         'Ministry of Justice "Youth Justice',
                         ' Indicator Report", 2021.'))

Labels in {ggplot2}

• xlab() and ylab() set the axis titles.

• Use NULL to remove the title.

  ylab <- ylab("Number of Offenders (per 10,000 population)")

• annotate() adds a geom.

  annotation <- annotate("text", label="COVID ", x=covid$x, 
                         y=Inf, hjust=1, vjust=1.5) 

Attention

• Titles and/or captions may provide top-down goals and tasks that direct attention.

Attention

• Titles and/or captions may provide top-down goals and tasks that direct attention.

Summary

• Text is an essential component of any data visualisation.

• We can use text data symbols to represent data values with great accuracy, but only for a small number of data values.

• We can use text labels to represent much more complex and abstract information, including metadata and data statistics.

• Labels are a very effective way to direct attention.

Exercise

Exercise

• Can you see anything wrong with this data visualisation?

  New Zealand Listener April 27-May 3 2024.