290 likes | 376 Views
Lecture 15: Interaction. March 28, 2013 COMP 150-2 Visualization. Admin. Update on Ben Shapiro’s data Group Assignment Review Incentivize projects? April 25 schedule? Invite department (and get pizzas) Thoughts on group projects…. Exercise.
E N D
Lecture 15:Interaction March 28, 2013 COMP 150-2Visualization
Admin • Update on Ben Shapiro’s data • Group Assignment Review • Incentivize projects? • April 25 schedule? • Invite department (and get pizzas) • Thoughts on group projects…
Exercise • Name all the types of tasks that you’d like your visualization to do
Analytic Activity in Information Visualization • Amar, Eagan, Stasko (2005) • Retrieve value • Filter • Compute derived value • Find extremum • Sort • Determine range • Characterize distribution • Find anomalies • Cluster • Correlate
Analytic Activity in Information Visualization • Amar, Eagan, Stasko (2005) • Retrieve value • What are the values of attributes X, Y, Z in the data points A, B, C? • Filter • Which funds under-performed the S&P 500 last year? • Compute derived value • What is the average income of CS grad students? • Find extremum • Which car has the highest MPG? • Sort • Order the cars by horse power
Analytic Activity in Information Visualization • Amar, Eagan, Stasko (2005) • Determine range • What is the length of this film? Who are the actors in this movie? • Characterize distribution • What is the age distribution of shoppers who purchase cars with 40+ MPG? • Find anomalies • Who are the outliers? • Cluster • Which cars are similar to each other in MPG, horse power, and price? • Correlate • Is there a relationship between horse power and MPG?
Different Directions and Possibilities • Animated control • http://www.nytimes.com/interactive/2009/11/06/business/economy/unemployment-lines.html • Hypothesis testing • http://www.lasvegassun.com/gambling-addiction/slotmachine/ • Algorithm to generate 1 image • http://flowingdata.com/2011/03/02/german-defense-ministers-plagiarized-phd-dissertation-visualized/ • http://internetcensus2012.bitbucket.org/paper.html • http://moviegalaxies.com/ • Tell a story • http://www.guardian.co.uk/world/interactive/2013/feb/12/state-of-the-union-reading-level • http://www.datapointed.net/visualizations/color/men-women-color-names-d3/ • Be Epic! Use visualization to show something cool… • http://flowingdata.com/2010/12/22/epic-animation-in-google-docs/ • http://salavon.com/SpecialMoments/SpecialMoments.shtml • http://www.nytimes.com/interactive/2013/02/22/sunday-review/the-consensus-candidate.html • http://flowingdata.com/2009/05/14/pixel-city-computer-generated-city/
Shneiderman’s Mantra • Ben Shneiderman, “The Eyes Have It: A Task by Data Type Taxonomy for Information Visualization”, 1996 IEEE Computer Society Press • “Overview first, zoom and filter, then details-on-demand”
Interaction • What is visualization without interaction? • What is the role of interaction in visualization? • Is it • visualization + interaction? • interaction + visualization?
Shneiderman’s Mantra • Overview first, zoom and filter, then details-on-demand • Overview • Visualize the data in its entirety • Zoom and filter • Let the user select and focus on the important stuff • Details on demand • Show what the user selected
Zoom and Filter • What is the user really doing when they perform “zoom” or “filter”? • Interaction == • “direct manipulation and instantaneous change” (Becker et al. 1987) • “the communication between user and the system” (Dix et al. 1998)
“Little Brother” • “Interaction is rarely the main focus of research efforts in the field, essentially making it the ‘little brother’ of [information visualization]” (Yi et al. InfoVis 2007) • As shown by Shneiderman, visualization == • Visual representation • Interaction
“Little Brother” • In most visualization courses, • Visual representation is the focus • Interaction receives little attention • But interaction is just as important if not more so! • Lots of opportunities because few people are thinking about it
Research • Designing completely new visual representations is hard • The domain is well covered • But given the same visualization, different interactions can allow different explorations, analyses, and discoveries • This is because interaction makes a visualization come “alive” • otherwise visualization is just a drawing on a piece of paper
Exercise: • What’s wrong with a (static) stacked graph? • http://www.meandeviation.com/dancing-histograms/hist.html
Technically… • What makes a visualization interactive? • In computer graphics, 12 frames per second is thought to be interactive because the animation will appear “smooth” to most people • 12 fps == 0.08 second per frame • Cognitively, • < 0.1 second : animation, continuity • < 1 second: system response, conversation break • < 10 second: cognitive response
Interaction as Selection • In most interactive visualization systems, we think of the interaction as the ability to select or filter some data items • Shneiderman’s “zoom and filter” • What does that mean technically?
Database Background • SQL Query. Example: • SELECT person • FROMdatasetTable • WHERE (weight > 130) AND • (weight < 180) AND • (height > 5.8) AND • (height < 6)
Database and Visualization Height = 6 Height = 5.8 weight = 130 weight = 180 • This is the equivalent of a selection box in a scatter plot visualization
Database and Visualization • For a specific type of interaction (e.g. select, filter, zoom, etc.), there is often a 1:1 mapping between the interactions and database queries • Homefinder Example
Database and Visualization • There is a nice mapping between each visual element with an attribute in the database • SELECT latitude, longitude • FROMhomesTable • WHERE(distance > 5) AND • (distance < 10) AND • (price > 100,000) AND • (price < 300,000) AND • (bedrooms > 3) AND • (garage = TRUE) AND • (fireplace = FALSE)
Query Relaxation • “Generalized Selection via Interactive Query Relaxation” (Heer et al. CHI 2008) • http://vis.berkeley.edu/papers/generalized_selection/
Database Strength and Weakness • Strengths: • Generalizable • Separates data from visualization cleanly • SQL is a very stable language that is very complete for interacting with data stored in relational databases • Weaknesses: • Dynamic query construction is tricky • DB overhead • Speed is a problem • Network speed • DB searching speed • Parsing DB returned tuples
Conjunctive Form • A selection / filtering / zooming interaction == database query • It can also be thought of as a conjunctive form • !(A1 V A2) ^ A3 V (A4 V A5 ^ A6) V … • where A1 could be the clause (price > 200)
Conjunctive Form • Very similar to SQL query, but can be built in software • Has some of the similar limitations as SQL in that the clauses need to be pre-built • Such that a data attribute needs to be “hard-coded” to a visual element • Consider a slider that connects to two or more attributes (http://gravis.cs.unibas.ch/publications/2007/VIS07_Smith.pdf)