GT CS 7450 - Multivariate Visual Representations

Unformatted text preview:

Multivariate Visual Representations CS 4460 7450 Information Visualization Jan 28 2010 John Stasko Agenda General representation techniques for multivariate 3 variables per data case But not lots of variables yet Spring 2010 CS 4460 7450 2 1 Quick Quiz Spring 2010 CS 4460 7450 How Many Variables 3 Revisit Data sets of dimensions 1 2 3 are common Number of variables per class 1 Univariate data 2 Bivariate data 3 Trivariate data 3 Hypervariate data Focus Today Spring 2010 CS 4460 7450 4 2 Earlier We examined a number of tried and true techniques visualizations for presenting multivariate typically 3 data sets Hinted at how to go above 3 dimensions Spring 2010 CS 4460 7450 5 Representations Some standard ways for low d data Tukey box plot low Middle 50 high 7 Mean 5 3 0 20 1 Spring 2010 CS 4460 7450 6 3 Hypervariate Data How about 4 to 20 or so variables for instance Lower dimensional hypervariate data Much higher dimensions next week Many data sets fall into this category Spring 2010 CS 4460 7450 7 More Dimensions Fundamentally we have 2 geometric position display dimensions For data sets with 2 variables we must project data down to 2D Come up with visual mapping that locates each dimension into 2D plane Computer graphics 3D 2D projections Spring 2010 CS 4460 7450 8 4 Wait a Second A spreadsheet already does that Each variable is positioned into a column Data cases in rows This is a projection mapping What about some other techniques Already seen a couple Spring 2010 CS 4460 7450 9 Revisit Multiple Views 1 Give each variable its own display 1 2 3 4 5 A 4 6 5 2 3 B 1 3 7 6 4 C 8 4 2 3 5 D 3 2 4 1 1 2 E 5 1 3 5 7 3 4 5 A B C D E Spring 2010 CS 4460 7450 10 5 Revisit Scatterplot Matrix Represent each possible pair of variables in their own 2 D scatterplot Spring 2010 CS 4460 7450 11 Chernoff Faces Encode different variables values in characteristics of human face length MPG cylinders width weight horsepower year Spring 2010 CS 4460 7450 12 6 Examples Cute applets http www cs uchicago edu wiseman chernoff http hesketh com schampeo projects Faces chernoff html Spring 2010 CS 4460 7450 13 Table Lens Spreadsheet is certainly one hypervariate data presentation Idea Make the text more visual and symbolic Just leverage basic bar chart idea Rao Card CHI 94 Spring 2010 CS 4460 7450 14 7 Visual Mapping Change quantitative values to bars Spring 2010 CS 4460 7450 15 CS 4460 7450 16 Tricky Part What do you do for nominal data Spring 2010 8 Instantiation Spring 2010 CS 4460 7450 17 CS 4460 7450 18 Details Focus on item s while showing the context Spring 2010 9 See It http www open video org details php videoid 8304 Spring 2010 Video CS 4460 7450 19 FOCUS Feature Oriented Catalog User Interface Leverages spreadsheet metaphor again Items in columns attributes in rows Uses bars and other representations for attribute values Spenke Beilken Berlage UIST 96 Spring 2010 CS 4460 7450 20 10 Spring 2010 CS 4460 7450 21 Characteristics Can sort on any attribute row Focus on an attribute value show only cases having that value by doubleclicking on it Can type in queries on different attributes to limit what is presented too Spring 2010 CS 4460 7450 22 11 Limit by Query Spring 2010 CS 4460 7450 23 Manifestation InfoZoom Commercial product to be demo ed coming up Spring 2010 CS 4460 7450 24 12 Categorical data How about multivariate categorical data Students Gender Female male Eye color Brown blue green hazel Hair color Black red brown blonde gray Home country USA China Italy India Spring 2010 CS 4460 7450 25 CS 4460 7450 26 Mosaic Plot Spring 2010 13 Mosaic Plot Women Spring 2010 Men CS 4460 7450 27 Mosaic Plot Brown Hazel Green Blue Women Spring 2010 Men CS 4460 7450 28 14 Mosaic Plot Black Red Brown Blond Brown Hazel Green Blue Women Spring 2010 Men CS 4460 7450 29 Intermission Readings reactions HW 1 recap Grade percentages Spring 2010 CS 4460 7450 30 15 Attr Explorer Discuss What is at the heart of the technique Spence Tweedie Inter w Computers 98 Spring 2010 CS 4460 7450 31 Characteristics Multiple histogram views one per attribute like trellis Each data case represented by a square Square is positioned relative to that case s value on that attribute Selecting case in one view lights it up in others Query sliders for narrowing Use shading to indicate level of query match darkest for full match Spring 2010 CS 4460 7450 32 16 Features Cost Attribute histogram All objects on all attribute scales 30K 100K Cost Interaction with attributes limits 100K 30K Spring 2010 CS 4460 7450 33 Features Inter relations between attributes brushing Cost 30K Bedroom 100K 1 5 Journey time 1h Spring 2010 30h CS 4460 7450 34 17 Features Color encoded sensitivity Cost 30K 1h Spring 2010 Journey time Bedroom 5 100K1 30h CS 4460 7450 35 Attribute Explorer Video http www open video org details php videoid 8162 Spring 2010 CS 4460 7450 36 18 Summary Summary Attribute histogram Attribute relationship Sensitivity information Especially useful in zero hits situations or when you are not familiar with the data at all Limitations Limits on the number of attributes Spring 2010 CS 4460 7450 37 MultiNav Each different attribute is placed in a different row Sort the values of each row Thus a particular item is not just in one column Want to support browsing Lanning et al AVI 00 Spring 2010 CS 4460 7450 38 19 Interface Spring 2010 CS 4460 7450 39 Alternate UI Can slide the values in a row horizontally A particular data case then can be lined up in one column but the rows are pushed unequally left and right Spring 2010 CS 4460 7450 40 20 Attributes as Sliding Rods Spring 2010 CS 4460 7450 41 Information Seeking Dialog Spring 2010 CS 4460 7450 42 21 Instantiation Spring 2010 CS 4460 7450 Demo 43 Limitations Number of cases horizontal space Nominal textual attributes don t work quite as well Spring 2010 CS 4460 7450 44 22 Dust Magnet Altogether different metaphor Data cases represented as small bits of iron dust Different attributes given physical manifestation as magnets Interact with objects to explore data Yi Melton Stasko Jacko Info Vis 05 Spring 2010 CS 4460 7450 45 CS 4460 7450 46 Interface Spring 2010 23 Interaction Iron bits data are drawn toward magents attributes proportional to that data element s value in that attribute Higher values attracted more strongly All magnets present on display affect position of all dust Individual power of magnets can be changed Dust s color and size can connected to attributes as well Spring 2010 CS


View Full Document

GT CS 7450 - Multivariate Visual Representations

Documents in this Course
Animation

Animation

23 pages

Load more
Download Multivariate Visual Representations
Our administrator received your request to download this document. We will send you the file to your email shortly.
Loading Unlocking...
Login

Join to view Multivariate Visual Representations and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view Multivariate Visual Representations 2 2 and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?