Download - Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Transcript
Page 1: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Information Visualization

Dianna XuDepartment of Computer Science

Bryn Mawr College

Page 2: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Big Data

• The amount of available data is quickly outpacing our ability to understand and use it in meaning full ways.

• The eventual solution to big data is likely really good black boxes.

• Black boxes only give answers if you know right questions to ask.

Page 3: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Visualization and Explorative Data Analysis

• Visualization allows one to view things in global context.

• Increase the likelihood of spotting unexpected trends and perspectives.

• Allows new questions to be asked.• The life cycle of explorative data analysis with visualization is likely an iterative process

Page 4: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Visualization Types

• Scientific visualization– numerical, high precision and has relative simple spatial relationships

– accurate representation of numerical proportions and realistic rendering of physical properties

• Data visualization– social/economical data– highly categorical and have strong association with physical locations/coordinates on the map 

Page 5: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Information Visualization

• Arbitrary or complex relationships with no clearly prescribed spatial representation choices

• Abstract representation techniques• Communication is an important goal• infoviz versus “data journalism”

Page 6: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

The Visualization Process

• Acquire• Parse• Filter• Mine• Represent• Refine • Interact

Page 7: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Mapping Numbers

• How to represent numbers with visual primitives so that the relative values of numbers can be quickly deduced from visual cues

• Example with simple data set– 255 integers valued 0‐99– collected on twitter as random numbers

Page 8: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Mapping Numbers

Page 9: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Mapping Numbers

Page 10: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Information Dimensions

• A fundamental challenge of visualization design is to add more information dimensions without introducing clutter

• Display is 2D• Going 3D is not a solution

Page 11: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Time Series

Page 12: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Stacked Graph

Page 13: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Heat Map

Page 14: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Birthdays

Page 15: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Proportional Symbols

Page 16: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Twitter Frequencies

Page 17: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

World Cloud

Page 18: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Packing

• Randomized placement allowing overlap• Randomized greedy algorithm

– fit the largest tiles first at randomly chosen location

– if there is overlap, try again

• Exhaustive search• Exhaustive search guided by space filling curve

– Spiral, Peano, Hilbert, etc

Page 19: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Spiral

Page 20: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Intersection via Pixel Buffer

Page 21: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Map‐based

Page 22: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Coordinate Conversions

• Translate physical coordinates to drawing coordinates

• Geographical coordinates• Modern world map empolying a Universal Transverse Mercator (UTM) projection

• What if your data comes with names of countries or states?

Page 23: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

A Choropleth Map

• Thematic maps require that you color an entire area (state) with a particular color

• Need polygonal outlines of a state• A US political map of SVG format contains coordinates as an XML file

• Combine with Google Maps API to create interactive map overlay

Page 24: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

States as Polygons

Page 25: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Google Map Overlay

Page 26: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Foursquares Check‐ins

Page 27: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued
Page 28: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Parallel Coordinates

• High dimensional numerical data• PNC Christmas Price Index

Page 29: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Parallel Coordinates

Page 30: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Brushing

Page 31: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Citeology

Page 32: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Parallel Sets

Page 33: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

TreeMap

Page 34: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

More Levels

Page 35: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Radial

Page 36: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Tree 

Page 37: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Radial Tree

Page 38: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Hyperbolic Tree

Page 39: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Focus+Context

Page 40: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Network/Graph‐based

Page 41: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Chord Diagrams

Page 42: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Circular Layouts

• Enjoyed great popularity lately• We are drawn to circles

– symmetry– aesthetics– perceived novelty

• Not always justified by increased readability• Human visual perception supports only rough comparison of areas and angles

Page 43: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Naming Names

Page 44: Information Visualization• Increase the likelihood of spotting unexpected ... numbers can be quickly deduced from visual cues • Example with simple data set – 255 integers valued

Data Analysis

• Unstructured data• A lot of it

• Dimension reduction• Cluster analysis• Statistical methods• Network models and optimization• Graph theory