Critique on Visualization by James Boyle and David Reichert
by Do Yeun Kim
Link to the visualizations
Real Data
Interactive Data Visualization
Abstract
The data used for this project is called the
Student Performance Data Set . This data contains 33 attributes
including the school, sex, age, absence, and alcohol consumption of the
total of 649 students. Of all possible questions, David and James were
primarily interested in visualizing the relationship between primary
caretaker and the secondary school of the students (visualization 1),
relationship between absence and academic performance of the students based
on the age of the student (visualization 2), and how skewed the grades
are according to age (visualization 2). Visualization 1 uses
stacked bar graphs that can transition into a regular bar graph, while
visualization 2 uses bar graphs that have 2 y-axes.
Stacked to Grouped Bar Charts
The Graph
This visualization is supposed to show the demographics of primary
caretakers in the two Portugese secondary schools of interest. JB and DR
achieves this by implementing a stacked/grouped bar chart, which shows
"Percent of Students" when stacked and "Number of Students" when
grouped, where the three options - "Other", "Father", and "Mother" - are
delineated by different shades of blue, according to the legend, and the
name and number of students of the secondary students are provided in the
x-axis.
The Good
The visualization is quite neat and easy on eyes, and the buttons, along
with the texts, do afford the user to click on them and explore the
interactivity of the graphs. Using both stacked and grouped bar graph
gives us accurate representation of the data, as difference in the sample
size from each school can cause misunderstanding of the data if only one
of the two were provided. Also, the transition is VERY cool.
The Not-so-Good
We need numbers. While y-axis does give us a general idea, lack of labels
definitely reduce clarity of the values. I also believe that the question
being asked is too mundane and narrow. So what if the primary caretaker
demographics are different between the schools? What does that tell us
about student performances? Why should I care about that? These do not
answer any of these more interesting questions, although they do show us
what JB and DR intended to show us.
Two y-axes Charts
The Graph
There are two bar graphs, one for each class - Math or Portugese - taken,
that have two y-variables mapped to age of the student, with two y-axes to
accommodate for the two y-variables. There are three buttons that allow the
user to prompt changes between displays of: "Average Grades & Absences",
"Maxim Grades & Absences", and " Above Average Population & Percentage." The
two y-variables are of different colors.
The Good
Interactivity is highly afforded, with the buttons with labels. The bars are
labeled so that we do not need to "guesstimate" the values that are
visualized. The bars are also arranged so that it is clear as to which bars
belong to which age.
The Not-so-Good
A two y-axes graph is by nature confusing, even though this visualization
attempts to counter the shortcoming with labelling and color coding. Another
thing is that it is not quite clear as to what question these graphs are
optimal answer. They show the trend of final grade and absences (or above
average and their percentages) for each age. They show the overall
qualitative tendency, but nothing specific. With the lack of specific
question being answered, I can only think that it would have been more
interesting to see the relationship between Final Grade and Absences
rather than simply have them mapped.
Final Verdict
The implementations are good. They got the data, took certain attributes, and
mapped them, with interactive functions and cool visuals. However, that is
where the merit of the project seems to end. These visualizations do not seem
to be answering any interesting questions. They simply take some attributes,
then map them to a somewhat ambiguous and more encompassing variable.
However, the graphs are indeed interesting and fun to look at, and I believe
that the implementation of the idea they had was good. If they had better
questions - which they do have them noted in
their
Brainstorming - the visualizations would have been much better.
Back to the Main Page