Data visualization and Presentation

What do you want to extract?
gendata.m
%% displaying numbers is not very helpful
%% displaying as an image is nicer (image, imagesc)
%% sorting first can reveal patterns

color maps

color maps
-- emphasize different things
-- continuity
-- if may be printed in b/w should work that way too

colormap info
In matlab you can make your own colormap -- just define a 3 column matrix and you can use it as a colormap. You can also use the colormap editor colormapeditor
e.g.

gmap = [ 1 0 0 ; .8 .1 .1; .7 .1 .1; .2 .1 .1; 0 .5 .5; 0 1 1]
colormap(gmap)
colorbar
rgbplot(gmap) %%% gives another view of the colormap
load('topo.mat','topo','topomap1');
rgbplot(topomap1)
%%%% (type earthmap.m)
What do you think of the use of these colormaps?

Are you plotting the whole picture?

(challenger disaster) click on missed opportunities here
click on context here

Case Study: What is bad/good about these displays of data?

what is wrong with this? (go to Feb 8 posting )
From Feb 8 posting What is better here?
What could still be improved?
http://www.cnn.com/ELECTION/2008/calculator/index.html
What is better in this format? (cartograms)
What if you just wanted to assess how the candidates were doing? What type of plot would you use?

If you wanted to show trends in the data

Visual representations should reflect effect size

click on Lie Factor and then Goosed up Graphics
Hinton Diagrams size (area vs length) hinton diagrams
hintonw([AA(Idx==1,:); AA(Idx==2,:); AA(Idx==3,:)]);
hintonc([AA(Idx==1,:); AA(Idx==2,:); AA(Idx==3,:)]);
colorbar;

Other useful references

http://www.stat.columbia.edu/~gelman/blog/ Gelman also has a book on classroom demonstrations for statistics classes: http://www.amazon.com/Teaching-Statistics-Tricks-Andrew-Gelman/dp/019857224 -- (http://www.edwardtufte.com/tufte/books_vdqi) is a book on best practices for visually representing data. http://www.amazon.com/How-Lie-Statistics-Darrell-Huff/dp/0393310728