CS838 Week 3_2 Vis

Assignment4: work in pair. (Jim Hill) : vis critique

==================

On mappings and visual encodings: importance of purpose. 
 - dealing with math and mapping in non-standard sense
 - "task bravery": if you know your task, you can dare making non-standard attempts
   => is it intended for a short route? long distance?

----------------
  
Data     	====> 	Variables 	       ====> 	visual attributes
   	mapping		visual encoding
	projection		design
Choosing encodings: refer to Munzner
's Vis. book chapter (Fig.27.5 adapted from Mackinlay)
 - networks: connection as data... categorical? quantitative?
  => bit of all. 
 - What encoding options are better for making comparisons?
  => least perceivable differences... position? color? (no, easily deceivable)


*John Snow's study: encoding position into position
position data ==> x,y ==> position
 
*What if we use position by color ? 
 - tying to other variables, such as # of houses by color
  => discarding position information: if not going into position-specific queries.
 - 'natural map' into scatterplot: less effective in quantitative exploration

* LineDrive case: 
 - dumps position encoding, uses connection ('sequence of instructions')
 - Have to be sure, to dare this approach: "no one will ask that (which I dumped)"
  limited resources vs rich information: constant compromise
  
Distortion: why is it good/bad? 
distance, orientation, shape, area
 - good: less cluttering, etc
 - bad: properties don't get preserved (angle, ratio, distance, continuity, invertable, etc)

 e.g. 3D sphere ->2D plain mercator map: giving up area accuracy to preserving continuity
 e.g. Fisheye view. log plots. 3D pie charts(bad vis). 
 image-retargetry: distorting without letting know the viewer