Please disable Adblockers and enable JavaScript for domain CEWebS.cs.univie.ac.at! We have NO ADS, but they may interfere with some of our course material.
AnswersQ7
- Ad p. 240 and generally on SPLOMS: How do I make the proper interpretation of a plot as a scatterplot or a SPLOM clear, as their goals are different and so must be their interpretation?
- Question is not clear to me -- SPLOMs are made up of many scatterplots. The interpretation of one scatterplot doesn't change?!
- When using one of the described reduction design choices interactively: How much impact can one have regarding there will be a window system runnin on the user's computer? Can we influence it? Do we even regard it when we visualize some data, or is that done in another 'field' (i.e. not part of the visualization process)?
- Again, not clear about the question: what concern do you have about the 'window system' on a persons computer? The Vis app is just like any other. Under special circumstances you can make certain assumptions (immersive environment, etc.)
- Why when a dataset has only one key, then it is straightforward to use that key to separate into one region per item?
- What else do you suggest to do with the key? Since spatial channel is the most important one, it is "straight forward" to use that one key with space.
- What is a recursive subdivision?
- You take the whole and (sub)divide it into 2 (or 3 or more) pieces. Then you *recursively* repeat this procedure for each partition.
- Why partitioning choice can accommodate a much larger number of views and layering not?
- I hope this became clear in class.
- Partitioning: Is there any data with specific characteristics that cannot be parititoned?
- not to my knowledge.
- Trellis: Is Trellis only suited for multidimensional tables with 3 categorical key atrributes and 1 qualitative attribute, as it is presented in the table on the p. 208 or it could be used for more complex data?
- could be used for more complex data as also explained in class.
- Beim Thema Cartographic Layering ist mir die Definition nicht so ganz klar. Warum stelle ich zwei Landkarten, eine in Farbe und eine in grau dar, wenn die Landkarte in Farbe übersichtlich ist?
- Was ist hier nicht klar? "The greyscale view shows that each layer uses a different range in the luminance channel, providing luminance contrast."
- Bei Superimposed line charts wird erwähnt, dass es bei einer geringen Anzahl von Daten eine gute Visualisierung bieten kann, jedoch bei einer großen Anzahl nicht. Kann man in dem Fall zum Beispiel nicht den Zeitslot vergrößern um damit quasi eine bessere Übersicht zu ermöglichen?
- Was heisst hier "Zeitslot"?? Ueber Animationen haben wir auch gestern geredet.
- Beim Beispiel HIVE wird erwähnt, dass es zum Beispiel verwendet werden kann um Eigenschaften von Daten darzustellen. In dem Fall Transaktionen in London mit Attributen wie Residence type oder dem Hauspreis. Wird das im Zusammenhang mit einer Landkarte verwendet oder wie soll es dann visuell nachvollziehbar sein?
- Nein -- HIVE benutzt keine Landkarten. Die Nutzer des Systems kennen London gut. Landkarten sind problematisch in vielerlei Hinsicht.
- Is there a difference between HIVE and a TreeMap? Or is 'The HIVE system' just a fancy name for an interactive TreeMap?
- HIVE ist ein system/tool. Es benutzt als als Methode Treemaps. Das sind zwei Verschiedene Konzepte. Man sagt auch nicht ein "Toyota ist ein fancy name für Auto".
- In the three design choices, partition, superimpose and change, is change really it's own choice? What will change, if you do not superimpose or partition?
- you can have a single view, show all data and only change the visual encoding (e.g. colour map). Good example were the maps of Florida where we simply changed the colour map.
- If data can be partitioned in multiple ways, is there a good strategy to select which views should be side by side to maximize pattern recognition/knowledge increase?
- Hmm, in general -- no. But it makes sense to have things "that belong" close to each other. E.g. small multiples where the different multiples change over time: in that case similar time steps should be next to each other. With other words, try to "respect the data" as much as you can.
- Looking on the superimpose layers I would like to know how is it going to be decided which view or which layer to be used and who makes the design decision?
- YOU! (the vis designer)
- It it OK when I use a big Number of items in multiple line charts and only change the width to see more of the values. Changing the width I am changing also the Y-Aray and so it is possible to see the difference between the values.
- you mean the line width? Line width can work well as a highlighting tool, i.e. whichever line the mouse hovers over becomes thicker. That's a good strategy.
- Why there is no linkage of different views according to the share navigation on page 231?
- good question.
- "If a glyph and a view can´t be divided very well, why these two different names are used?" and "In 10.3.1 nennt glyph.Was meint man hier mit glyph?"
- glyph and view are different concepts. Syntactically they have to do with size. A glyph is small and we typically use tons of them at once. Views are large, provide lots of detail and we have very few.
- I didn´t understand what the Hierarchical edge bundles really is and for what it can be used. Is there an easier explanation? Will it be discussed in the lecture?
- yes, we did
- I don't really understand what it is that figure 10.3 should point out exactly. Even with the explanation for the figure, I can't make much use of it. Why is the Morris site in the third row an outlier, but not the Wasoca row in the first?
- lots of details in class.
- What is the blue line in figure 10.10? Is this supposed to show these two values in the different windows? Is this then basically the same as "linked highlighting", which gets mentioned on the next page?
- discussed in class.
- On page 233, Tamara mentions something called bivariate sequential-sequential color encoding. What does that mean? (I tried googling it, but the search didn't render any usable results, so I consider this a legitimate question.)
- It was discussed on colormaps, see Fig 9.1: bivariat is clear (two colours), sequential-sequential alludes to 2D colormap.
- What are the ways to make distinguishable layers?
- That's too general -- the whole book talks about it, esp. chapter 5.
- What is the difference between dynamic layer and static layer?
- Static -- remains the same, cannot change, can only add or delete. Dynamic -- I can change it through mouse clicks. see Fig 10.9
- Why ist superimpose very important by visualisation?
- Why not? I don't understand the question.
- Is a solar plot a good choice for aggregating data? It doesn't seem pretty readable to me, or at least not the one pictured in the book or the top results that google gives me. Also, the controls and the resulting effects (i.e. circle size) don't seem intuitive for me.
- This is chapter 11 already! It might work in particular applications … but I am not a big fan either. :)
- We already learned that displaying a measure series by its statistical key figures can get really problematic by looking at Anscombe's Quartet, which shows 4 completely different measurement series with the same statistical key figures. How can one decide, when it is a good choice to use item aggregation by key figures?In the book it is mentioned, that this is a really though choice, but I didn't catch any real strategies to make this decision a bit more systematic, rather than intuitive.
- tricky. Depends on questions you'd like to answer, the amount of data you need to go through and what what level of detail is important to solve a task. Perhaps summary statistics are good enough to filter out a number of irrelevant data ...
- If I, for example, use a map of Vienna to visualize some data (let's say public wlan hotspots) for the average user - how can I decide, if it is a better option to make an "unrestricted zoom" possible, or to force the user to be satisfied with, let's say 7 layer of detail? Or is this rather a question of personal taste? In particular, how big can the problem of computational power on the client side become if I allow free camera movement?
- I think the layer idea is about semantics on the one hand -- which items belong together and can't / shouldn't be separated. Further, things in the same layer usually share the same visual encoding. I hope this helps.
- All the visualizations that are shown in this chapter seem to need some kind of explanation to read. Is this not something that should be considered aswell when talking about who is going to read these?
- Often times the explanations are due to the fact that we don't know the context in which they were designed and what "visual literacy" the users of these tools already have. So, no, I wouldn't be worried! :)
- When talking about the space that is required shouldnt the factor of where someone is going to read it? For example nowadays everything seems to be moving towarrds mobile devices, which would make different partitions of the same dataset more interesting and other facetes less.
- sure, but even in 20 years we will have small screens AND big screens. Maybe not LCD 17' monitors, but your living room wall. I don't think that large screens will disappear in favour of small screens. However, doing proper vis for small screens and proper interaction / navigation for touch screens and vis has not been well understood yet. Room for you to make contributions!
- When sharing data across views in a mobile device, wouldnt other types of sharing be usefull. For example sharing values to compare in a different view (that you cannot see on the same screen)
- Don't exactly follow, but see answer above -- more research is needed. Vis is not "done" yet.
- Dynamic Layers - Should the interaction with the user regarding data visualization be more discussed in Visualization or HCI?
- Not sure what you are asking. But "Visualization" without interaction is like a car without gas (or a battery) :) It's pretty to look at, but not useful :)
- Can animations still be the right choice, if they are very good designed? (especially because of the ability to be used as an eye-catcher)
- One cannot answer this question as general as such -- it depends! "Catching your eye" and working with a tool for hours to solve a problem are two different beasts!
- When i program a visualisation software which is capable to interact with the user regarding the visualisation. isn't that a danger to my intended visualization design? the user could just superimpose with too much layers and make the software responsible for the bad visualisation. How could I avoid this?
- Well, don't give the users too many options. You, as the programmer, are the all-powerful Zen master. If you think three layers and no more -- thy wish shall be granted (well, if you have the sufficient programming skills) :)
- Am Anfang des Kapitels ist eine Grafik die die Verhältnisse zwischen "encoding" und "data" darstellt. Am Ende des Kapitels wird diese Grafik wieder angezeigt aber ich habe sie leider auch dann noch nicht verstanden. Ich wüsste gerne was die Grafik aussagt.
- Ich hoffe dies ist in der Vorlesung klar geworden!
- Bei Grafik 10.6 steht "layering works well because of the luminance contrast between the elements on different layer". Ich wüsste jetzt gerne was hier gemeint ist. Bei der farbigen Grafik sehe ich zwar unterschiedliche Farben aber ist hier wirklich die Helligkeit auch entscheidend?
- Der Kontrast -- ja, sehr! Wir hatten da einige Beispiele in der Vorlesung!? Wenn immer noch unklar, bitte mit mir reden.
- What are usual dataset partioning strageties?
- I hope this became clear in the lecture.
- Is it still a superimposed layer, if the layer can be turned on and off? If yes, it is possible to create more than two or three layers (as mentionend in the book) and just turn it off/on, isn't it?
- sure. just like photoshop!?
- A view is redundant when it shares all the data and the encoding with another view. So if i view a bone for example from different directions, just to get a better overview, it would be redundant, or is the point of view a encoding of data too?
- yes -- camera angle is considered part of the visual encoding.
- […] schwierig zu verstehen ist […] das Beispiel zur „Principal Components Analysis“, vielleicht könnten Sie es daher in der VU noch einmal erläutern.
- Ich hoffe es hat geholfen!
Critique / Additional Notes
- Figure 10.8, compound graph: I don't feel, that this is either a good example for spacial partition nor for superimposition. Nothing really overlaps and there is only one view. Also, can you make out anything in this picture? To me it looks like chaos.
- Dagegen habe ich das Example: Hierarchical Edge Bundles nicht ganz verstanden und ich finde die Darstellung auch etwas unübersichtlich und verwirrend.
- Noch unübersichtlicher finde ich nur noch das Example: Cerebral. Hier kann man in der Ansicht eigentlich fast gar nichts mehr erkennen und das was gezeigt werden soll ist nicht wirklich zu sehen.
- […] It gets even worse with the HIVE example. Unfortunately, I don't understand what it is supposed to visualize, probably because I don't understand the visualization technique itself (it looks like a multi-matrix plot to me, but I might be completely wrong). So, because I don't understand the visualization itself, it doesn't really help me understand "Partition By Values", even though the concept doesn't seem that hard.
- What is the blue line in figure 10.10? Is this supposed to show these two values in the different windows? Is this then basically the same as "linked highlighting", which gets mentioned on the next page?- yes, these are highlights.
- Missing References:There are quite a few references to 'Figure ??', altough for many of them the images are already there, eg p.221,Context-Preserving Visual Links- yes! sorry!
- Possible Errors:Page 204, bottom:3Vocabulary note: The word glyph <MISSING>is<\MISSING> used very ambiguously in <MAYBE_REMOVE>the<\MAYBE_REMOVE> vis literature.
- I found the entrance to facets, right at the beginning of chapter 10 (10.1 The Big Picture) very hard to read. The first sentence is "This chapter covers wether and how to facet data across multiple views." what ARE facets? Also figure 10.1 didn't made clear to me what she (the author) was trying to tell. Then we have "A view is a single window, a contiguous region in which visually encoded data is shown on the display." Ok, that is clear but what has that todo with facets? Again, what are facets? It wasn't even clear by re-reading the previous chapters. To me it looks like a gap between chapters 9 and 10, I think the transition between chapter 9 and 10 is way too hard.
- After being a little bit disappointed by not knowing what facets actually are, we come to chapter 10.2 - Why Facet? Yep, that didn't helped me either on my quest to understand what facets are. The following chapters where clearly and understandibly written (especially figure 10.2 with the example was very good), but unlike in the previous chapters, I had trouble finding the big picture here.
- Die vielen Beispiele und Bilder sind mir auch diesmal wieder positiv aufgefallen, etwas schwierig zu verstehen ist meiner Ansicht nach jedoch das Beispiel zur „Principal Components Analysis“
Letzte Änderung: 02.04.2014, 20:42 | 2665 Worte