Validation of Visualizations CS 4390/5390 Data Visualization Shirley Moore, Instructor September 24,
Why Validate? Vis design space is huge and most visualizations are ineffective. Validate choices throught design and implementation process so as not to have to tear up and redo 2
Four Levels of Vis Design 3
Domain Situation Target users Their domain of interest Their data Their questions Each domain has its own vocabulary for describing its data and questions. Usually some existing workflow Example: Computational biologist using genomic sequence data to ask questions about the genetic source of adaptivity in a species Vis designer needs to clearly understand users’ needs 4
Requirements Elicitation Outcome: Deatiled list of questions to be asked about the data Which is better? 1) What is the density of coverage and where are the gaps across a chromosome? OR 2) What is the genetic basis of disease? 5
Task and Data Abstraction Map domain-specific questions into abstract vis tasks such as browse, compare, summarize – This is an identification step. Choose the most appropriate data abstraction and transform original data if needed – This is a creative design step. 6
Encoding and Interaction Idioms Visual encoding idiom – create a picture of the data Interaction idiom – how users control and change what they see Make design decisions based on understanding of human abilities such as visual perception and memory 7
Algorithms Efficient implementation of visual encoding and interaction idioms Accuracy of data representation may also be an issue. May have choice of different algorithms – e.g., different volume rendering algorithms for creating images from MRI data 8
Threats and Downstream Validation 9
Validation Example Sizing the Horizon by Heer, Kong, and Agrawala –
Class Exercise 1 Write down questions to be answered by your Lab 2 visualization Interview a classmate about what questions they want answered about the data Revise your questions if needed 11
Class Exercise 2 Working with the same person you interviewed for the preceding exercise, share your What? Why? How? analysis for Lab 2 Validate whether your data and task abstractions match the questions from Exercise 1 12
Preparation for Next Class Prepare downstream validation tests for Lab 2 visualizations 13