Information Visualization

- Arranging tables
- Zero keys
- Scatterplots

- Some keys
- Barcharts
- Heatmaps

- Other axis orientations
- Sploms
- Parallel coordinates
- Navio

- Radial representations
- Pie charts
- Radar plots

Space is the most important channel.

- Quantitative data
**👉 Express** - Categorical data
**👉 Separate, order, align**

- Independent attribute
- Used as unique index to look up items
- Simple tables:
**one key** - Multidimensional tables:
**multiple keys**

- Dependent attribute, value of cell
- Classify arrangements by key count
- Zero, one, two, many...

**Express**values- Quantitative attributes
- No keys, only values
**Data**: two quantitative attributes**Mark**: points**Channels**: horizontal and vertical position**Tasks**: identify trends, outliers, distribution, correlation, clusters**Scalability**: hundreds of items

**Regions**: contiguous bounded areas distinct from each other- Using space to
**separate**(proximity) - Following expressiveness principle for categorical attributes
- Use ordered attribute to
**order**and**align**regions

- One key, one value
**Data**:- One categorical attribute, one quantitative attribute
**Mark**: lines**Channels:**- Length to express quantitative value
- Spatial regions: one per mark
- Separated horizontally, aligned vertically
- Ordered by quantitative attribute
- By label (alphabetical), by length attribute (data-driven)
**Task:**- Compare, lookup values
**Scalability**- Dozens to hundreds of levels for key attribute

- Two keys, one value
**Data:**- Two categorical attributes
- One quantitative attribute
**Mark:**vertical stack of line marks**Glyph:**composite object, internal structure from multiple marks**Channels:**length and color hue- Spatial regions: one per glyph
- Aligned: full glyph, lowest bar component
- Unaligned: other bar components
**Task:**part-to-whole relationship**Scalability**: few bars, few stacked

Interesting discussion from Robert Kosara's blog.

Make sure to read the comments.

- Labeled axis is critical
- Avoid cropping y-axis
- Include 0 at bottom left
- Or slope misleads

- Two keys, one value
- Data
- Two categorical attributes (gene, experimental condition)
- One quantitative attribute (expression levels)
- Marks: area
- Separate and align in 2D matrix
- Indexed by two categorical attributes
- Channels
- Color by quantitative attrib
- (Ordered diverging colormap)
- Task: find clusters, outliers
- Scalability
- One million items, hundreds of categorical levels, about ten quantitative attribute levels

- In addition
- Derived data
- Two cluster hierarchies
- Dendrogram
- Parent-child relationships in tree with connection line marks
- Leaves aligned so interior branch heights easy to compare
- Heatmap
- Marks (re-)ordered by cluster hierarchy traversal
- Task: assess quality of clusters found by automatic methods

- Rectilinear axes, point mark
- All possible pairs of axes
- Scalability
- One dozen attributes
- Dozens to hundreds of items
- Task:
- Summarize/explore

- Parallel axes, jagged line representing item
- Rectilinear axes, item as point
- Axis ordering is major challenge
- Scalability
- Dozens of attributes
- Hundreds of items
- Task:
- Summarize/explore

- Scatterplot matrix
- Positive correlation
- Diagonal low-to-high
- Negative correlation
- Diagonal high-to-low
- Uncorrelated: spread out

- Parallel coordinates
- Positive correlation
- Parallel line segments
- Negative correlation
- All segments cross at halfway point
- Uncorrelated
- Scattered crossings

- Rectilinear: scalability with regard to #axes
- Two axes best
- Three problematic
- Four or more impossible
- Parallel: unfamiliarity, training time

What about millions of points?

http://vis.stanford.edu/projects/immens/demo/splom/- Radial bar chart
- Radial axes, meet at central ring, line mark
- Star plot
- Radial axes, meet at central point, line mark
- Bar chart
- Rectilinear axes, aligned vertically
- Accuracy
- Length unaligned with radial
- Less accurate than aligned with rectilinear

- Task: comparison
- Scalability: similar to parallel coordinates
- Order of attributes matter

- Pie chart:
- Area marks with angle channel
- Accuracy: angle/area much less accurate than line length
- Polar area chart:
- Area marks with length channel
- More direct analog to bar charts
- Data:
- One categorical key attribute, 1 quantitative value attribute
- Task:
- Part-to-whole judgements

- Some empirical evidence that people respond to arc length
- Not angles
- Maybe also areas...?
- Donut charts no worse than pie charts

- Meta-points
- Redesign of paper figures in later blog post
- Violin plots good for analysis but too detailed for presentation
- Tamara's advice: still dubious for pie/donut charts
- Sometimes okay if just two attributes

- Task:
- Part-to-whole judgments
- Normalized stacked bar chart
- Stacked bar chart, normalized to full vertical height
- Single stacked bar equivalent to full pie
- High information density: requires narrow rectangle
- Pie chart
- Information density: requires large circle

- Rectilinear good for linear vs. nonlinear trends
- Radial good for cyclic patterns

- Perceptual limits
- Polar coordinate asymmetry
- Angles lower precision than lengths
- Frequently problematic
- Sometimes can be deliberately exploited!
- For two attributes of very unequal importance

- Data: text
- Text and one quantitative attribute per line
- Derived data:
- One pixel high line
- Length according to original
- Color line by attribute
- Scalability

- Arranging tables
- Zero keys
- Scatterplots

- Some keys
- Barcharts
- Heatmaps

- Other axis orientations
- Sploms
- Parallel coordinates
- Navio

- Radial representations
- Pie charts
- Radar plots