Comments on: Reading 2: Overview

By: lyalex

lyalex — Tue, 26 Jan 2010 19:39:03 +0000

I have read Tamara Munzner’s Visualization, Chapter 27. The chapter 27 gives us the basical ideas in visualization, such as four layers and encoding principles.
The most useful thing that I find is Figure 27.5, which lists the most efficient way for the three different kinds of data for human-beings to perceive. Another useful thing is the examples throughout section 27.8, which also give me a lot of thoughts for the template and the methodology which can implement my experimental results into more understandable ways. Basically, as an novice in the visualization field, Tamara’s chapter is a very good start for me to understand basic concepts of visualization.

By: Chaman Singh

Chaman Singh — Tue, 26 Jan 2010 15:39:39 +0000

I read Visualization ( Tamara Munzner ) and chapter-I ( Stuart Card ). Both
are introductory yet provides lots of insights into effective visualization
techniques.

The nice things about Stuart paper are Table 1.3 and 1.20 which succinctly
explains the concept of “Cognition Amplification”. But probably I disagree
with the authors that visualization amplifies cognition because it expands
working memory. I think it is the efficient utilization of limited memory
and not the expansion which enhances “Gestalt Organization” which is given
in Table 1.20.

Figure 1.26 is a great example to show that good visual structures ( and not
only the colors) are extremely important to understand hidden patterns in a given
dataset.

I couldn’t fully appreciate or understand one of the sentences in the paper
“Distortions can be roughly classified by what human perceives as invariant”.
Also not clear what they mean by the following sentence “Distortion is not
effective when the features or patterns of use to the user are distorted
in a way harmful to the task”.

Tamara paper has opinion against the efficacy of animation as a visualization
tool, which is quite surprising, as most people in scientific community
prefer animation rather than static images.

By: Nakho Kim

Nakho Kim — Tue, 26 Jan 2010 14:52:16 +0000

Like many others on this comment list, I’ve also read the Tufte piece and Munzner piece. It was a interesting pair because the two chapters seemed to be on both ends of a spectrum for “how to explain good visualisation”. While Tufte provides a philosophy of good visualisation, Munzner approaches from the specific technics up. Also the former explains the importance of showing more and complex datasets, the latter emphasizes reducing of the dataload.

Also, I liked Munzner’s four-step approach in choosing a visualization method, which was a nice complement (or rather, providing the grund principle for) the questions-based process of the Mizbee paper of Week 1.

By: dalbers

dalbers — Tue, 26 Jan 2010 14:47:56 +0000

Munzner’s chapter is very much based around the idea of teaching the reader what exactly visualization is. It spends a good deal of focus explaining the “vocabulary” of visualization, taking a look at the more technical aspects of developing visualizations. Even perceptual ideas such as color and layout are explained with a style geared toward a visualization scientist, focusing on the theory of what perceptual techniques will function better in a visualization as opposed to what will create a more aesthetic design.

Tufte’s reading, however, seems to characterize exactly what we were told about the author. Instead of a professional and scientific tone, passages such as that theorizing the stomach cancer rates in the northeast are based on the smoke fish by the Scandinavians living there or “A silly theory means a silly graphic” set the author up as being under a critical (and somewhat judgmental) tone, very much in contrast to Munzner’s more scientific and professional writing styles. However, Tufte’s inclusion of so much history of both visualization and the examples provided new insight into the significance and power of the techniques as well as giving great visual cues as to how to conduct effective design, even if the reader is somewhat biased by Tufte’s proclamation of a graphic being “the best ever drawn.” His analysis of technique, directed at the designer, is based on logical claims and present strong arguments on how to design effective visualizations.

By: faisal

faisal — Tue, 26 Jan 2010 14:46:09 +0000

I read cahpters from Munzner and Tufte’s book and also read through half of the Card et.al. book’s chapter.

These readings provided a good insight into the whole process involved in visualizing information. Specially, InfoVis paper explained well details about the principles involved in visualizing data sets. All details in Munzner paper are backed by some fairly intuitive examples. The Tufte’s design perspective presents visualization by explaining qualities and use of different graphics design. The focus seems to be more on understanding components of graphics design and how much information they can convey instead of explaining the whole process of visualization. I personally liked the InfoVis style of approaching visualization as that seems to be more systematic and intuitive. But, definitely both perspective are very important for visualization and there should exist some way to exploit information in both these perspectives.

The historical information presented in Tutfe’s readings were also interesting. Specially, the fact that many of the visualization primitives that are perhaps no-brainer now (e.g. time-series plots) didn’t come under use until late in eighteenth century. One more thing that caught my eye while reading Card et.al.’s book chapter was how bad visualization can be dangerous e.g. in case of accident of space shuttle Challenger. In this case, clutter and lack of suitable visual encoding for temperature data prevented O-ring manufacturers from seeing a clear pattern for bad launch temperature.

Overall, I think these readings were great in the sense that they motivates well the problem of visualization and provide good grounds for further readings and discussion.

By: aditya

aditya — Tue, 26 Jan 2010 14:07:30 +0000

I first read Card, Mackinlay, Schneiderman. I liked they way they introduced the notion of external cognition using a plethora of examples of daily life. The fact that we use vision to think, and, moreover, it is this use of external aids which gives us humans an edge and “makes us smart”. There were also plenty of subsequent examples which illustrated how appropriate visualization could “reveal the truth” and also reveal insights.
They also distinguished between scientific visualization (which uses physical data) and information visualization (which uses abstract data): they main difference (and challenge) being that nonphysical information has no obvious spatial mapping.
By stating the Principle of Selective Omission of Information and the Cost-of-Knowledge Characteristic function, they made the process of visualization more quantitative: perform abstraction of information, yet make sure access to useful information is not too costly.
I found the reference model for visualization to be quite instructive. Interestingly, though unsurprisingly, a lot of program analysis follow the same steps of going from raw data (concrete semantics) to more manageable visual structures (abstract semantics) using which understanding the property of interest becomes easier. In program analysis, there is a well-defined notion of when an abstraction is ‘safe’ or sound. It would be interesting to see whether such a notion has an analogue in information visualization.

Tufte’s chapter showed how great visualization could be in communicating data and showed-off the variety of ways in which data could be graphically visualized. He also illustrated how visualization could be misleading in certain situations. I found his principles of Graphical Excellence to be very instructive, especially his observation that “graphical excellence is nearly always multivariate.”

By: Shuang

Shuang — Tue, 26 Jan 2010 05:45:26 +0000

I am reading Tamara Munzner’s Visualization, Chapter 27 and Tuft’s. The chapter 27 focuses on the importance of visualization in science world. It presents extra information besides the automatic solution by machines.

The data types part, 27.2, explains how to make data visualized. It is interesting to know how to visualize all kinds of data, including quantitative, categorical ones and graphic data. This provides a clear way to view the data. Also, how to withdraw information from data is important, with data transformation skills. The chosen information will illustrate the essence of data. Figure 27.3 shows the nested structure of validation of visualization, which is the procedure of what a scientist follows.

It gives ideas of infovis, and explains how to visualize data clearly. Besides numbers, color, grid, text, etc., can be informatic to explore the data.

The Tuft’s provides a lot of examples and some of them are not the trivial summary of data. That leads me to think what would be the best figures to show when I have data. We should pay much attention on choosing figures for audience.

By: watkins

watkins — Tue, 26 Jan 2010 05:37:35 +0000

I read the Munzner chapter first, then the Tufte chapter, but I wish I would have read them in the opposite order. Tufte begins and ends his chapter with a list of criteria for “Graphical Excellence”. His chapter is focused on history and analyzing examples of “good” graphical representations, most of them hand drawn. However, he doesn’t provide any insight as to how to make a good graph, just what it is. Munzner’s chapter is not only mainly focused on methodology, but cites much more modern examples, making it a great followup to the Tufte reading. And, both papers touch on some similar ideas (like higher-than-spatial dimensional representation, glyphs, and the conclusion that there IS a difference between a good and bad visualization) in very different ways that complement each other well.

By: jeeyoung

jeeyoung — Tue, 26 Jan 2010 05:14:31 +0000

I read Tamara’s and Tufte’s.

The Tufte’s chapter has lots of examples which we might have seen some time. These examples give me a sense of what visualization aims to and how nontrivial to come up with great graphics.

Tamara’s introduction approaches visualization in different angle – visualization science. It is a nice introduction of infovis.

To me, “Visualization Scientist” perspective is to dissect a problem and follow the cascade of layers with validation while Tufte’s “Designer” perspective is rather spontaneous.

By: Michael Correll

Michael Correll — Tue, 26 Jan 2010 04:26:15 +0000

The Tufte reading was very refreshing: it really made a lot of sense to see the graph as an evolutionary design rather than an “obvious” method of presenting data that are related. I think there is sometimes the tendency to see graph-making as just something that mathematical people are naturally good at, despite all of the evidence to the contrary.

The notion that there is are right and wrong ways to present data seems to ring true to my experience. Somebody writing a paper will spend a lot of time refining their writing to make it clear for the intended audience, and yet the graphics are often left unconsidered.

When visualizations are good, they are really really good. When they are bad, they are very very bad.

The Card et al. reading just seemed to confirm this. The O-Ring graph from the Challenger really hammered home the point that there are very real consequences for getting this process wrong.