Masterclass creating business value with big data 

Datavisualisation

Maarten Lambrechts

Vlerick Business School

Brussels - February 20 2015

After this session you'll know

how

numbers

can talk more

efficiently

to

your & others

brains

Maarten Lambrechts?

Data & Multimedia journalist @Mediafin (De Tijd & L'Echo)

 

Bio-Engineer (KULEUVEN 2003)

 

Datavisualisation trainer & consultant

"As knowledge increases amongst mankind, and transactions multiply, it becomes more and more desirable to abbreviate and facilitate the modes of conveying information from one person to another, and from one individual to the many."

William Playfair, 1786

"Everything that can be expressed in numbers can be represented by lines."

John Playfair

East

West

North

South

Visualisations

 

answer questions

 

go beyond descriptive statistics

 

show things otherwise hidden

The secret?

preattentive processing

868888609462834719714449648
122286651978794222484779908
939340243123007392183390116
351684124962334339709749742

Attentative Processing

868888609462834719714449648
122286651978794222484779908
939340243123007392183390116
351684124962334339709749742

preAttentive Processing

Good visualisations use

Color

size

position

rotation

shape

enclosure

deliberately

Subtle

smart

Another Secret

Gestalt laws

Use the laws of

proximity

connectedness

similarity

enclosure

 

The basics

Ground rules for efficient visualizations

Evaluation

some simple datavisualisation 

EXAMPLES we can learn from

Evaluate on

  1. Correctness

  2. Readability

  3. Efficiency

  4. Use of text and numbers

  5. Use of color

  6. use of preattentive and gestalt features

NL

Re-evaluation with checklist

 

Which one is officialy the worst visualisation?

For more bad visualisations: viz.wtf

Datavisualization*

communicates Numbers

Faster

More efficient

More complete

aesthetically more pleasing

with More fun

*When done right

webcharts

USE

Tooltips

Filter

extra context

Switch view

Drill down

Sort

Animation

Deliberately

SUBTLE

smart

Tool:

Tableau

tableau

  • respects ground rules

  • Drag 'n drop

  • Many data connections

  • Sharing is easy

Demo

Where to move if you want to meet a rich and single Flemish woman?

tableau is not

  • a data management platform

  • a spreadsheet editor

  • Statistical modelling tool

Dashboards and macroscopes

why dashboards?

Question:

Monitoring

Anwser specific questions

  • Everything ok?
  • Processes ended?
  • Cause of problem?
  • Where to invest resources?

Exploration

Gain insights 

  • Who are our customers?
  • How are people using our webshop?
  • What are the patterns in our sales?

Monitoring

Anwser specific questions

  • Only show relevant data
  • User is most important
  • Should lead to action
  • Tool = dashboard

Exploration

Gain insights 

  • Show all the data
  • Data is most important
  • Should lead to insight
  • Tool = macroscope

10 design principles

for dashboards

8 also apply to macroscopes

user centric

NOT DATA CENTRIC

NOT DASHBOARD CENTRIC

Use the grid

Left top to bottom right

Gradual reveal

tooltips

Pop ups

drilldowns

1. overview

Dataviz mantra

2. zoom

3. details on demand

respect datastructure

Time series

Categories

Flows

Hierarchies

respect dataviz groundrules

Guide attention 

Color

Type

Sizes

Grid

Grouping

show context

Targets

Past time periods

Averages

Other statistics

avoid unnecessary precision

1.4k vs 1.458,54

USE labels, but limit them

Guide action

Actions to perform

People to contact

Evaluation

Let's Apply these principles to real life examples

ING Luxembourg's complaints dashboard

User centric

Use the grid

Gradual reveal

Dataviz mantra

Respect datastructure

RESPECT DATAVIZ GROUND RULES

GUIDE ATTENTION

SHOW CONTEXT

AVOID TOO MUCH PRECISION

GUIDE ACTION

 

Macroscopes

 

Belgian Car Fleet

Bibliography