Ch 7+8: Tables, Spatial Data Arrange not much about the ones in - PowerPoint PPT Presentation

News VAD Ch 7: Arrange Tables Arrange tables • clarification on artery vis Encode Axis Orientation Express Values Rectilinear Parallel Radial – diverging colormap since doctors care about high and low values Ch 7+8: Tables, Spatial Data Arrange • not much about the ones in the middle Express Separate Separate, Order, Align Regions – personal communication with Borkin, not clearly stated in paper Separate Order • second guest lecture today from Kosara Layout Density – vis for presentation (versus discovery/exploration) Tamara Munzner Dense Space-Filling Order Align • then continue with lecture/discussion Department of Computer Science Align – catch up on chapters, leave papers for Thu University of British Columbia • remember CPSC 547, Information Visualization Use – I have office hours on Tuesdays Day 8: 6 October 2015 1 Key 2 Keys 3 Keys Many Keys List Matrix Volume Recursive Subdivision – pitches are coming up Thu Oct 22 http://www.cs.ubc.ca/~tmm/courses/547-15 – start talking to me about project ideas! 2 3 4 Idiom: scatterplot Idiom: bar chart Keys and values Some keys: Categorical regions Tables 100 100 Express Values • key • express values • one key, one value 75 75 Attributes (columns) Separate Order Align 50 50 – independent attribute Items – quantitative attributes – data (rows) 25 25 – used as unique index to look up items • no keys, only values • 1 categ attrib, 1 quant attrib 0 0 Cell containing value – mark: lines – simple tables: 1 key – data – channels – multidimensional tables: multiple keys Multidimensional Table • 2 quant attribs Animal Type Animal Type • length to express quant value • regions : contiguous bounded areas distinct from each other • value – mark: points • spatial regions: one per mark – channels – using space to separate (proximity) – dependent attribute, value of cell – separated horizontally, aligned vertically Value in cell • horiz + vert position – following expressiveness principle for categorical attributes • classify arrangements by key count – ordered by quant attrib – tasks • use ordered attribute to order and align regions » by label (alphabetical), by length attrib (data-driven) – 0, 1, 2, many... • find trends, outliers, distribution, correlation, clusters – task 1 Key 2 Keys 3 Keys Many Keys – scalability Express Values 1 Key 2 Keys 3 Keys Many Keys • compare, lookup values List Matrix Volume Recursive Subdivision List Matrix Volume Recursive Subdivision • hundreds of items – scalability • dozens to hundreds of levels for key attrib [A layered grammar of graphics. Wickham. Journ. Computational and Graphical Statistics 19:1 (2010), 3–28.] 5 6 7 8 Idiom: stacked bar chart Idiom: streamgraph Idiom: line chart Choosing bar vs line charts 60 60 20 50 50 • one more key • generalized stacked graph • one key, one value • depends on type of key attrib 40 40 15 30 30 – data – emphasizing horizontal continuity – data – bar charts if categorical 20 20 10 10 10 • 2 categ attrib, 1 quant attrib • vs vertical items • 2 quant attribs – line charts if ordered 0 0 Female Male 5 Female Male [Stacked Graphs Geometry & Aesthetics. Byron and Wattenberg. – mark: vertical stack of line marks – data – mark: points • do not use line charts for IEEE Trans. Visualization and Computer Graphics (Proc. InfoVis 0 60 60 2008) 14(6): 1245–1252, (2008).] • glyph : composite object, internal structure from multiple marks • 1 categ key attrib (artist) • line connection marks between them 50 50 categorical key attribs 40 40 – channels • 1 ordered key attrib (time) – channels 30 30 – violates expressiveness principle [Using Visualization to Understand the 20 20 Year • 1 quant value attrib (counts) • length and color hue • aligned lengths to express quant value 10 10 Behavior of Computer Systems. Bosch. Ph.D. • implication of trend so strong that – derived data 0 0 • spatial regions: one per glyph thesis, Stanford Computer Science, 2001.] • separated and ordered by key attrib into horizontal regions 10-year-olds 12-year-olds 10-year-olds 12-year-olds it overrides semantics! – aligned: full glyph, lowest bar component • geometry: layers, where height encodes counts – task after [Bars and Lines: A Study of Graphic Communication. – “The more male a person is, the Zacks and Tversky. Memory and Cognition 27:6 (1999), – unaligned: other bar components taller he/she is” • 1 quant attrib (layer ordering) • find trend 1073–1079.] – task – scalability – connection marks emphasize ordering of items along key axis by explicitly showing relationship between • part-to-whole relationship one item and the next • hundreds of time keys – scalability • dozens to hundreds of artist keys • several to one dozen levels for stacked attrib – more than stacked bars, since most layers don’t extend across whole chart 9 10 11 12 Idiom: heatmap Idiom: cluster heatmap Idioms: scatterplot matrix, parallel coordinates Axis Orientation • scatterplot matrix (SPLOM) • two keys, one value • in addition Scatterplot Matrix Parallel Coordinates Math Physics Dance Drama Rectilinear Parallel Radial – rectilinear axes, point mark – data – derived data Math 100 – all possible pairs of axes • 2 categ attribs (gene, experimental condition) • 2 cluster hierarchies 90 80 Physics • 1 quant attrib (expression levels) – dendrogram – scalability 70 60 – marks: area • parent-child relationships in tree with connection line marks • one dozen attribs 50 Dance 40 • separate and align in 2D matrix • leaves aligned so interior branch heights easy to compare • dozens to hundreds of items 30 20 Drama – indexed by 2 categorical attributes – heatmap • parallel coordinates 10 Many Keys 1 Key 2 Keys 0 – channels Math Physics Dance Drama List Matrix Recursive Subdivision • marks (re-)ordered by cluster hierarchy traversal – parallel axes, jagged line representing item Table • color by quant attrib – rectilinear axes, item as point – (ordered diverging colormap) Math Physics Dance Drama • axis ordering is major challenge 85 95 70 65 – task 90 80 60 50 – scalability • find clusters, outliers 65 50 90 90 50 40 95 80 • dozens of attribs – scalability 40 60 80 90 • hundreds of items • 1M items, 100s of categ levels, ~10 quant attrib levels after [Visualization Course Figures. McGuffin, 2014. http://www.michaelmcguffin.com/courses/vis/] 13 14 15 16