syllabus‎ > ‎R: Intro‎ > ‎

Mokeresete-Kris-Motlhala

VISUALIZATION

Dimension and Item Count

Are there any new developments on techniques that would better work for high-dimentional datasets?

2D vs. 3D
It seems 2D representation of data is more useful than 3D. Are there any cases where 3D can be more useful in terms on data represetation?


POLARIS
This wasn't an easy one for me.

Using shape size for encoding can be a very useful representation if one wants to see the pattern for certain activity like sales over months, but is sometimes important for one to be able to see how much in terms of sales is that size of the shape used for encoding. If pattern or trends can be shown at the same time with figures represented by those same shapes then it might be useful. However, if drilling down is supported in such a way that if you point or click on any shape you see the real data that is represented, then that would be good.
Ooh.., I think the tooltip addresses this.

How does binning differ with categorization?

Polaris seems to be a very good tools for those who are learning to use it. The fact that it has unlimited undos and redos allows the user to explore and experiment different visualization options. You know you will always be able to correct you mistakes. That is a good feature

I just find polaris useful from the examples presented in paper. You can perform several statistical functions like Sum, Average, Count in combination with visualization.


INTERACTION
I think what the writer said is true that interaction is as important as representation though there is not so much research done of interaction as compared to interaction. .
There is no way interaction can be divorced from representation.

Interaction in the context of InfoVis is more on users of the InfoVis system getting information from the system than them (user) feeding the system with information.


Categories

Categorization of techniques was done basing on the intent, what the user want to achieve in performing performing the interaction.
  1. Select
  2. Explore
  3. Reconfigure
  4. Encode
  5. Abstract/Elaborate
  6. Filter
  7. Connect
Conclusion
It is difficult to create categories of interaction techniques that are clear and comprehensive. The categories we proposed are based on our own perspective on interaction in Infovis and, thus, inherently debatable.

Some techniques are difficult to classify and do not quite fit into any one of the categories.

Compare category was not included because is implicit in other categories, and I agree.
Subpages (1): Tableau Projects
Comments