Exercise 7 | Theory of Data Graphics II

Max Pellert

IS 616: Large Scale Data Analysis and Visualization

Friendly

Words are spelled out, mysterious and elaborate encoding avoided

Words run from left to right, the usual direction for reading occidental languages

Little messages help explain data

Elaborately encoded shadings, crosshatching, and colors are avoided; instead, labels are placed on the graphic itself; no legend is required

Graphic attracts viewer, provokes curiosity

Friendly

Colors, if used, are chosen so that the color-deficient and color-blind (5 to 10% of viewers) can make sense of the graphic (blue can be distinguished from other colors by most color-deficient people)

Type is clear, precise, modest; lettering may be done by hand

Type is upper-and-lower case, with serifs

Unfriendly

Abbreviations abound, requiring the viewer to sort through text to decode abbreviations

Words run vertically, particularly along the Y-axis; words run in several different directions

Graphic is cryptic, requires repeated references to scattered text

Obscure codings require going back and forth between legend and graphic

Unfriendly

Graphic is repellent, filled with chartjunk

Design insensitive to color-deficient viewers; red and green used for essential contrasts

Type is clotted, overbearing

Type is all capitals, sans serif

Aspect Ratios

When we are working with data graphics, we are usually quite free to choose the dimensions of our graphic (height and width)

In the case of vector graphics, we can choose arbitrary (also very large) dimensions without loss of quality

For bitmap graphics, there are limits after which quality noticeable decreases and we don’t want that

An important question concerns the ratio of width to height, the aspect ratio

Which aspect ratio to choose for one specific data graphic?

Line Width

Heavier lines should be a data measure

As an example consider a time series plot:

Coding

Increase the ink-data ratio of Anscombes Quartett

anscombe <- datasets::anscombe

# earlier function
# create_plot <- function(dataset_x,dataset_y,size_points=4,size_text=21){
#   ggplot(anscombe,
#          aes({{ dataset_x }},{{ dataset_y }})) +
#   geom_point(
#     size = size_points) +
#   geom_smooth(method="lm", se=F, fullrange = TRUE,
#               color="darkgrey") +
#   scale_x_continuous(
#     breaks = seq(0,20,2)) +
#   scale_y_continuous(
#     breaks = seq(0,14,2)) +
#   expand_limits(x = c(0,20), y = c(0,14)) +
#   labs(x = deparse(substitute(dataset_x)),
#        y = deparse(substitute(dataset_y))) +
#   theme_bw() +
#   theme(text=element_text(size=size_text))
# }

p0 <- ggplot(anscombe, aes(x1,y1))

p0

p01 <- p0 + geom_point(size = 2.5)

p01

p02 <- p01 + scale_y_continuous(breaks = c(5,10)) +
  expand_limits(y = c(0,14))

p02

p03 <- p02 + scale_x_continuous(labels = c("",10,"",20),
                                breaks=c(5,10,15,20),
                                expand = c(0, 0)) +
  expand_limits(x = c(0,20))

p03

p03 <- p02 + scale_x_continuous(labels = c("",10,"",20),
                                breaks=c(5,10,15,20),
                                expand = c(0, 0)) +
  expand_limits(x = c(0,20))

p03

p04 <- p03 + labs(x = "", y = "") + theme_classic()

p04

p05 <- p04 + annotate("text", x = 2, y = 13, size=10, family="Times", label = "I")

p05

p06 <- p05 + labs(x = "", y = "") + theme(text=element_text(size=16),
      axis.text.x = element_text(hjust = 0.5),
      axis.line = element_line(colour = 'black', linewidth = 0.6),
      axis.ticks = element_line(colour = "black", linewidth = 0.5),
      axis.ticks.length=unit(.15, "cm"),
      plot.margin=unit(c(.2,.5,.2,.2),"cm"))

p06

p1 <- ggplot(anscombe,
       aes(x1,y1)) +
geom_point(
  size = 2.5) +
# geom_smooth(method="lm", se=F, fullrange = TRUE,
#             color="darkgrey") +
annotate("text", x = 2, y = 13, size=10, family="Times", label = "I") +
scale_x_continuous(
  labels = c("",10,"",20), breaks=c(5,10,15,20), expand = c(0, 0)) +
scale_y_continuous(
  breaks = c(5,10)) +
expand_limits(x = c(0,20), y = c(0,14)) +
labs(x = "",
     y = "") +
theme_classic() +
# theme_bw() +
theme(text=element_text(size=16),
      axis.line = element_line(colour = 'black', linewidth = 0.6),
      axis.ticks = element_line(colour = "black", linewidth = 0.5),
      axis.ticks.length=unit(.15, "cm"),
      plot.margin=unit(c(.2,.5,.2,.2),"cm"))

p2 <- ggplot(anscombe,
       aes(x2,y2)) +
geom_point(
  size = 2.5) +
annotate("text", x = 2, y = 13, size=10, family="Times", label = "II") +
scale_x_continuous(
  labels = c("","10","","20"), breaks=c(5,10,15,20), expand = c(0, 0)) +
scale_y_continuous(
  breaks = c(5,10)) +
expand_limits(x = c(0,20), y = c(0,14)) +
labs(x = "",
     y = "") +
theme_classic() +
theme(text=element_text(size=16),
      axis.line = element_line(colour = 'black', linewidth = 0.6),
      axis.ticks = element_line(colour = "black", linewidth = 0.5),
      axis.ticks.length=unit(.15, "cm"),
      axis.text.y = element_text(colour = "white"),
      axis.text.x = element_text(colour = "white"),
      plot.margin=unit(c(.2,.5,.2,.2),"cm")) 
  

p3 <- ggplot(anscombe,
       aes(x3,y3)) +
geom_point(
  size = 2.5) +
....

p1 + p2 + p3 + p4 + plot_layout(ncol = 2)

https://patchwork.data-imaginist.com/

Bonus

Add rugplots to the axes…

…to detect outliers and label outliers

Small multiples

Could you provide an alternative solution not using patchwork or a similar package?

Think about rearranging the data (and of using facets in ggplot2 for example)

For another, Python solution: Vega-Altaire

https://iliatimofeev.github.io/altair-viz.github.io/gallery/anscombe_plot.html

import altair as alt
from vega_datasets import data

anscombe = data.anscombe()

anscombe
##    Series   X      Y
## 0       I  10   8.04
## 1       I   8   6.95
## 2       I  13   7.58
## 3       I   9   8.81
## 4       I  11   8.33
## 5       I  14   9.96
## 6       I   6   7.24
## 7       I   4   4.26
## 8       I  12  10.84
## 9       I   7   4.81
## 10      I   5   5.68
## 11     II  10   9.14
## 12     II   8   8.14
## 13     II  13   8.74
## 14     II   9   8.77
## 15     II  11   9.26
## 16     II  14   8.10
## 17     II   6   6.13
## 18     II   4   3.10
## 19     II  12   9.13
## 20     II   7   7.26
## 21     II   5   4.74
## 22    III  10   7.46
## 23    III   8   6.77
## 24    III  13  12.74
## 25    III   9   7.11
## 26    III  11   7.81
## 27    III  14   8.84
## 28    III   6   6.08
## 29    III   4   5.39
## 30    III  12   8.15
## 31    III   7   6.42
## 32    III   5   5.73
## 33     IV   8   6.58
## 34     IV   8   5.76
## 35     IV   8   7.71
## 36     IV   8   8.84
## 37     IV   8   8.47
## 38     IV   8   7.04
## 39     IV   8   5.25
## 40     IV  19  12.50
## 41     IV   8   5.56
## 42     IV   8   7.91
## 43     IV   8   6.89

alt.Chart(anscombe).mark_circle().encode(
    alt.X('X', scale=alt.Scale(zero=False)),
    alt.Y('Y', scale=alt.Scale(zero=False)),
    column='Series'
).properties(
    width=200,
    height=200
)

Until next week…

Get the data…

https://waf.cs.illinois.edu/discovery/berkeley.csv

https://discovery.cs.illinois.edu/dataset/berkeley/

Until next time

Visualize the Berkeley data as an informative graphic (and a table and possible a combination of both), investigating admission rates by gender (following the design principles discussed in the course so far)

What do you find? Think about possible reasons for your findings

Until next time

Now, create small multiples split up by the study program applicants applied to (Variable “Major”) and take a look at admission rates by gender again for each of the majors

What do you find now? What could be the reasons for your earlier findings?

What other informative aspects of the data are there to be uncovered? Visualize them with techniques that you deem adequate