Data Visualization

The Visualization Continuum

Charts & Dashboards (communicate changes in measure)
Finding Relationships (Explore dimension relationships for meaning)
Physical Map

Goal: facilitate compensation of data in a quick way. User needs to understand it AND make decisions.

Data Visualization Definition

•Thedisplay of abstract information (S. Few) (For Sense-making, For Communication)
•Presentation of quantitative data to:
–Facilitate consumption and comprehension
–Illuminate difference and trends
–Support decision making

Data is useful, matters, and deserves a response.

Why is data visualization effective?

“Data visualization shifts the balance between perception and cognition to take fuller advantage of the brains abilities”
We process visual information faster than verbal information.
The human visual system is optimized to see differences (size, shape, contrast) DMD

Data Visualization in Practice

“Data visualization is only successful to the degree that it encodes information in a manner that our eyes can discern and our brains can understand”And we can make decisionsbased on it!
Useful= Informs + matters + deserves a response

Success Criteria for Visualization

•Clearly indicates the nature of relationships
•Represents the quantities accurately
•Makes it easy to compare the quantities
•Makes it easy to see ranked order in the values
•Makes obvious how people should use the information to make decisions
–Aka it has “Meaning”

Operational Principles

1.Define/research users goals (HF analysis)
2.Fix the data first!
3. Highlight the important stuff
• Use pre-attentive attributes like bold and color
• No more than 10% of the content
4. Eliminate distractions
• Tufte’s data to ink ratio applies!
5. Provide clear information hierarchy
6. Provide filters for manipulation (questions)

The critical Human Factors Challenge

  1. Assess information users really need
  2. Understand their goals
  3. Select and aggregate the data properly

Anatomy of Quantitative Data

•Measures(the raw metrics)–Sales, Profit, Search Views, User count, Votes
–Not all numbers are measures (Emp. ID, SSN, etc)
•Dimensions (organize the data)
–Hierarchical •Time •Geography
–Categorical •Divisions of  Comp
•Relationship/Filters (hide & show the data meaning)
–Sorting (male versus female consumers)
–Cluster data (Mobile versus Web product usage)

What is the story to be told? (by design to the user when they look at data)

1) Variations within Measures
–Normal, non-normal distributions
–Homogeneity or heterogeneity of data
–Means and medians
2) Relationships between Measures
–Co-variance, etc.
–Correlation & causality (statistics)
3) Variations within Dimensions(Categories)
–Part to whole
–Count, sum & average
4) Variations across Time
5) Variations across Space (geography)

Measures can be raw, normalized (units per sq ft), or summarized (statistics like median or subtotals). What determines how the measures are designed to be displayed? Goals.


Dimensions Organize the data

• Hierarchical or Categorical
• Exact Schema (IA)  –Time & Geography
• Natural dimensions (given scalar values like Money, Headcount)
•Synthetic dimensions (designed, dynamic)
–Corporation, Line of Business, Sales regions, Distribution method (Rail, Ship, etc)
*don’t compare to business goals because those change all the time

The big three data design tasks are COMPARISON, CATEGORIZATION, and NORMALIZATION.

Critical for UX design:

1)Always compare to something!
2)Always show the trend direction and magnitude!
3)Sometime up is good, sometimes it is bad
4)90% of all information consumption uses case can be solved with this pattern

How to slice up data? Clustering, Conversions, Normalization (to a comparison or baseline)

Visualization -relationships within charts

Dimension relationships:
•Nominal (East, West)
•Ordinal (Best, Worst)
•Interval (days, months)
•Hierarchical (States > counties)
Measure relationships:

Always ask first: can you show just the data or do you need a chart?

The power of charts
The visual objects in chart can encode both Measures (height = sales) and Dimensions (color =region). A table can’t.


Visual Design

  1. No Gradient background
  2. No meaningless use of color
  3. Data to ink ratio
  4. Aspect ratio matters (max slope of 45 degrees)
  5. Variable order matters


Interaction with visual info

• Comparing (time series, reference value, rate of change)
• Sorting
• Filtering
• Highlighting (“brushing” –synchronized color selection in dashboards)
• Re-visualizing (change chart type)
• Re-expressing (change units, e.g. dollars to euros)
• Re-scaling (time, days vs. months)
• Zooming and panning
• Detail on demand (hover over point)
• Drill to detail
• Annotate
• Bookmark
• Aggregating (e.g. sales by product by region by year, pivot)
• Adding measures (raw or summarized, statistics & forecasts)



Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s