Introduction to Data Visualization Review

Lec 1

what: The use of computer-supported, interactive, visual representations of data to amplify cognition; Computer-based visualization systems provide visual representations of datasets intended to help people carry out some tasks more effectively.
Visualization models
borrows from Perceptual and cognitive science. models for developing visualization tools.
“Value of Visualization”: a typical visualization process maps from data and control parameters to images.
“Readings in Information Visualization: Using Vision to Think”: Information Visualization Reference Model.
Why need visulization
The age of big data, provide a way to simplify big data. Data is huge, complex, and heterogeneous.
Power of Visual Cognition: 1. Statistics sometimes give wrong answers e.g. Anscombe’s quartet: mean, sample variance, correlation. can not see difference from set of data, visualize can see 2. Eyes can cheat. Optical illusion:Use color, light and patterns to create images that can be deceptive or misleading to our brains.

Lec 2

multiple dimension data: tabular data, A multidimensional table enables data to be modeled and viewed in multiple dimensions. A data cube is created from a subset of attributes; Dimensions are a fact that defines a data cube. roll-up/drill-down; slicing/dicing
slicing: trans to table(low-dim); dicing:small cube; For detailed, hierauchical: roll-up 聚类; drill-down 分解
data attribute: categorical; Ordered (Ordinal, quantitative, ording direction);
Data cleaning: struc/format; missing or corrupted values; Why 80%? 1. Remove duplicate or irrelevant observations 2. Fix structural errors 3. Filter unwanted outliers 4. Handle missing data 5. Validate and QA
Difference between cleaning/transfor: Data cleaning is the process that removes data that does not belong in your dataset. Data transformation is the process of converting data from one format or structure into another.
join tables: 1. find common types of data 2. multiplication 3. delete rows that doesn’t match the key

Lec 3

preattentive processing: The ability of the low-level human visual system to rapidly identify certain basic visual properties.
-Without need for focus attention -Tasks completed in less than 200 to 250 ms -parallel processing
(need a sequential search, no conjunction)
Limited set of visual elements: Form 形状 and Hue 颜色; orientation, length, flicker, closure, 3D depthcue, 3D orientation…

coding schemes: luminance (for details); colors (dissimilarity); integral: 2++ features together; separable: each dimension;
Redundant coding: seperable dimensions for same thing. Interference: one visual attributes performance on the other.
color and shape are the most seperable visual components. color+position: good design.
Gestalt Theory (Six): Proximity: 整体大于部分之和 1+1>2 临近性：靠近的为一组 Similarity continuity: 排成一线的物体，引导视线 Closure: 部分缺失靠想象 Symmetry: balanced/perceptually pleasing figure-ground

Lec 4

4 coordinate systems: Data coordinates: range of x and y View volume coordinates Normalized view coordinates: scaling to [0,1] Screen coordinates. DataMapping: viewvolume to screen 计算：1. 写初始数据范围 2. 写结果范围 3. 移原点矩阵 4. scale screen矩阵 5. 扩大到 (180,120) 6. 移动原点到 (100，100)

Lec 5 dark red 700nm to violet 400nm

color space: RGB CMYK RYB HSL Lab
color map: sequential: 关注极大值; categorical; Diverging: 三个值一样重要;

Lec 6

Tufte’s rules: Use graphics; Let the data speak; Use labels; Avoid chartjunk; Utilize data-ink ratio; Utilize micro/macro; Use small multiples;
Not to lie with data visualization: Truncated Y-Axis; Cumulative graphs; Ignoring conventions; Inconsistent scales; Size & volume encoding;
Lie factor= size of effect shown in the graphic / size of effect in the data. In [0.95, 1.05]
= {(5.3-0.6) / 0.6} / {(27.5-18) / 18} = 783% / 53% = 14.8

Lec 8

univariate, bivariate, trivariate; High-dimensional data visualization: Visual mapping: multiple views, scatterplot matrix, iconic representation, table lens, parallel coordinates

Lec 9

The role of interaction in information visulization: Select; Explore; reconfigure; encode; abstract/elaborate; filter; connect;

Lec 10

Mercator projection: conformal:local shapes are reserved; cylindrical; can be used for navigation;