13/10/2022

Effective Data Visualization with Tableau

Effective Data Visualization Notes Copyright © 2022 by Krisolis LTD, Dublin, Ireland. All rights reserved.
No part of this publication may be reproduced, transmitted, in any form or by any means, electronic,
mechanical, or otherwise, without the prior written permission of the publisher, Krisolis.

1
 13/10/2022

Introductions

▪ After attending this course, learners will

understand how to design and build data
visualisations that can be used to effectively
communicate actionable insights from data.
At Course ▪ Learners will be able to:
Completion - Move from business questions to great data
visualisations and beyond
- Apply the fundamentals of data visualisation to
create informative charts
- Choose the right visualisation type for the job at
hand

2
 13/10/2022

- Design and develop basic dashboards in python

that people will love to use by doing the
following:
 Creating the following types of charts - cross tabs,
pie and bar charts, geographic maps,, highlight
At Course tables, tree maps and scatter plots
Completion  Creating dashboards

Start End Topic

09:15 10:45 Fundamentals of Data Viz
10:45 11:00 Coffee Break
11:00 12:30 Introduction to Tableau

Schedule 12:30 13:30 Lunch Break

Day 1 13:30 14:30 Perception Theory
14:30 14:45 Coffee Break
14:45 15:45 Preparing Data in Tableau
15:45 16:00 Coffee Break
16:00 17:00 Building a Simple Dashboard

3
 13/10/2022

Start End Topic

09:15 10:45 The Data Visualization Zoo
10:45 11:00 Coffee Break
11:00 12:30 The Data Visualization Zoo

Schedule 12:30 13:30 Lunch Break

Day 2 13:30 14:30 Enhancing Visualisations
14:30 14:45 Coffee Break
14:45 15:45 Building Dashboards
15:45 16:00 Coffee Break
16:00 17:00 Workshop

‘The function of a dashboard is

to communicate critical
information to your audience in
a way they can understand,
delivered when and where they
need the information’
Juice Analytics

Juice Analytics "A Guide to Creating Dashboards People Love to Use”

www.juiceanalytics.com/s/Guide_to_Dashboard_Design.pdf

4
 13/10/2022

Visualization Fundamentals

10

In this
section we
will take
inspiration
from some
great data
visualisations
to get us
started!

11

5
 13/10/2022

Hans Rosling (Legend!)

Hans Rosling TED Talk: The Best Stats You’ve Ever Seen
www.ted.com/talks/hans_rosling_shows_the_best_stats_you_ve_ever_seen.html

12

Hans Rosling

https://www.youtube.com/watch?v=hVimVzgtD6w
13

6
 13/10/2022

Purpose of
Data 1. To tell a story and communicate
Visualisation insights in an effective manner

14

1854 London Cholera Epidemic

▪ On the evening of August 31, 1854
Cholera broke out in the Broad Street
area of central London
▪ John Snow, a London doctor, was
asked to investigate
▪ Visualisation was key to making the
authorities believe what he found out

Adapted from “The Visual Display of Quantitative Information”, Graphics Press USA, 2001
www.edwardtufte.com/tufte/books_visex

15

7
 13/10/2022

1854 London Cholera Epidemic

16

1854 London Cholera Epidemic

17

8
 13/10/2022

21

Purpose of
Data 2. To solve a problem
Visualisation

22

9
 13/10/2022

23

London Underground Map

London
Underground
map circa 1930

www.vam.ac.uk/vastatic/microsites/1331_modernism/highlights_19.html

24

10
 13/10/2022

London Underground Map

Draughtsman Harry
Beck, thought, "If you're
going underground, why
do you need bother about
geography? …
Connections are the
thing" and designed a
new map.

www.vam.ac.uk/vastatic/microsites/1331_modernism/highlights_19.html

25

London Underground Map

29

11
 13/10/2022

Other City
Metro Maps
now use the
same format

30

Purpose of
Data 3. To display data in the most
Visualisation informative form, making it easy to
uncover trends

31

12
 13/10/2022

Anscombe’s Quartet
▪ Frank Anscombe was an English statistician
- Made significant contributions in design
of experiments and regression methods
- An early proponent of
statistical computing
and stressed that
"a computer should
make both calculations
and graphs”

35

Purpose of
Data 4. To uncover hidden trends in data
Visualisation that sometimes can be hidden by
simply looking at statistical output

36

13
 13/10/2022

▪ The key points made in this section were

- Visualisations can be used to illuminate
complex data and data analysis for a wide,
general audience
Section - Simple looking visualisations usually have a
huge amount of work underlying them
Takeaways
- Sometimes visualising the data can present
problems in a new way which allows new
solutions
- Sometimes visualising data shows insights
that numeric analysis hides

37

Workshop

▪ Now we will do work

in groups to assess
the effectiveness of a
number of
visualisations

38

14
 13/10/2022

Introduction to Charting
with Python

39

▪ Python is an open source, general purpose

programming language

▪ Developed out of the ABC language by the

Dutch programmer (and Python's Benevolent
What Is
Dictator for Life) Guido van Rossum
Python?
▪ Release history
- First release (version 0.9.0) was in 1991
- Python 2.0 released in 2000
 Now (finally) discontinued
- Python 3.0 released in 2008
 Version 3.10 released in late 2021
www.python.org

40

15
 13/10/2022

IEEE
Spectrum
Programming
Language
Ranking

https://spectrum.ieee.org/top-programming-languages/

41

▪ Open source and well supported by freely

available tools
▪ Clean, concise, unambiguous syntax
▪ Supports a variety of programming
Why paradigms
Python? - Simple scripts
- Object-oriented programming
- Web applications
- Interactive notebooks
▪ Strong built-in and third party library support
▪ Strong online community support

42

16
 13/10/2022

THE SCIPY
ECOSYSTEM

43

Python As A Collection Of Libraries

▪ Python has a large built-in standard library that includes
support for database connections, networking, handling
regular expressions, …
▪ One of the biggest attractions of Python is the fact that is
easily extended through the development of libraries (or
packages)
▪ Libraries can be written by anyone, but are available centrally
through installers such as pip
▪ Just about everything you
can imagine is implemented
in some Python package!
▪ See https://pypi.org

44

17
 13/10/2022

Jupyter
display Pandas SciPy

pickle Python Matplotlib

Language Python
Core
NumPy

math Python
Standard
Library
seaborn

scikit-learn plotly

45

▪ The pip tool is the recommended tool for

installing Python packages
▪ It can be run from the command line and
makes managing packages easy…
Installing $ pip install SomePackage
Packages

$ pip install --upgrade SomePackage

$ pip uninstall SomePackage

46

18
 13/10/2022

▪ SciPy (pronounced “Sigh Pie”) is a

Python-based ecosystem of open-
source software for mathematics,
science, and engineering containing
the following key packages:
- NumPy: Provides numerical array
The SciPy types and mathematical operations
Ecosystem - SciPy: Core Python library used for
scientific computing and technical
computing
- Pandas: Easy to use data structures
and operations for manipulating
them
- Matplotlib: Publication-quality 2D
plotting library
The SciPy Ecosystem:
www.scipy.org
www.numpy.org
pandas.pydata.org
www,matplotlib.org

47

▪ NumPy is a Python library that provides

functionality for numerical
computing containing:
- A powerful N-dimensional array object – i.e.
vectors and matrices
NumPy - Efficient functions for manipulating arrays
- Useful linear algebra, Fourier transform, and
random number generation capabilities

▪ Besides its obvious scientific uses, NumPy is

also used as an efficient multi-dimensional
container for generic data in many data
analytics projects
www.numpy.org

48

19
 13/10/2022

▪ SciPy is a collection of mathematical

algorithms and convenience
functions built on the NumPy
extension of Python providing
- Functionality for manipulating and visualising
data, building on NumPy arrays
SciPy
- Sophisticated optimisation algorithms and
statistical functions
- Tools for integrating C/C++ and Fortran code

www.scipy.org

49

▪ Pandas is a Python library that provides

flexible, easy-to-use data structures
and data analysis tools
- In many ways Pandas provides a Python
equivalent of the data analysis and
manipulation tools available in the R
Pandas programming language.
- Pandas builds on packages from the SciPy
Ecosystem: NumPy, SciPy, and Matplotlib
- Pandas works well as part of the Jupyter
Notebook environment

pandas.pydata.org

50

20
 13/10/2022

▪ Matplotlib is a comprehensive library

for creating static, animated, and
interactive visualisations in Python
- Allows us to create publication quality plots and
export to many common file formats
Matplotlib - Support for interactive figures that can zoom,
pan, and update.
- Customisable visual styles and layouts
- Visualisations can be directly embedded
in Jupyter Notebooks and other visual front-
ends

www.matplotlib.org

51

▪ Seaborn is a Python data

visualisation library based on
Matplotlib. It provides a high-level interface for
drawing attractive and informative statistical
graphics.
- Integrates closely with Pandas
- Seaborn plotting functions can operate on
Seaborn DataFrames and NumPy arrays containing whole
datasets
- Supports a wide range of visualisation types
- Offers many styles and customisation options
- Generally produces "nicer" graphics than basic
Matplotlib plots

import seaborn as sns

seaborn.pydata.org

52

21
 13/10/2022

▪ When importing packages from the SciPy

ecosystem the following package aliases are
recommended
▪ We don’t have to do this, but everybody does!
import numpy as np
Package import pandas as pd
Aliases import scipy as sp
import matplotlib as mpl
import matplotlib.pyplot as plt

53

BASIC DATA VISUALISATIONS

WITH PANDAS SERIES

54

22
 13/10/2022

▪ iPython is the engine underlying Jupyter

Notebooks
▪ iPython has a set of predefined magic
commands that you can call with a command
iPython line style syntax for doing various system-
Magic level jobs. These are not part of the standard
Functions Python language!
▪ Line magic commands are prefixed with the
% character and work much like OS
command-line calls taking an argument the
rest of the line
▪ Cell magic commands are prefixed with a
double %% and take all the lines in a cell as
iPython magic functions:
https://ipython.readthedocs.io/e
an argument.
n/stable/interactive/magics.html

55

▪ Useful magic functions:

- %run runs a python script (.py file)
- %cd change the current working directory
iPython - %colors swap between defined colour
Magic schemes:
Functions NoColor, Linux, LightBG
- %precision set floating point precision for
pretty
printing
- %system run a system shell command

iPython magic functions:

https://ipython.readthedocs.io/e
n/stable/interactive/magics.html

56

23
 13/10/2022

- %%latex render cell contents as latex

- %%javascript interpret cell contents as
Javascript
iPython - %R executes a line of R code
Magic - %matplotlib sets up Matplotlib options to
work interactively. Adding inline causes
Functions
Matplotlib plots to be displayed in
Jupyter notebooks
%matplotlib inline

Always include this to make sure

your Matplotlib images appear in
your notebook!
iPython magic functions:
https://ipython.readthedocs.io/e
n/stable/interactive/magics.html

57

▪ The Pandas package includes convenient

functions to generate basic plots and charts
▪ The Pandas package wraps calls to
visualisation functions from the Matplotlib
package
Pandas Plots ▪ We will cover the following graphs
- Bar plots
- Pie charts
- Histograms
- Kernel plots
- Box plots
- Time series line plots

58

24
 13/10/2022

▪ The Pandas plot function is an easy way to

draw common plots from a Pandas series

pd.Series.plot(kind='line'
figsize=None,
Pandas Plots title=None,
grid=None,
legend=False,
xticks=None, yticks=None,
xlim=None, ylim=None,
colormap=None)
pandas.plot:
pandas.pydata.org/pandas-
docs/stable/generated/pandas.S
eries.plot.html

59

The kind of plot to

▪ The Pandas plot function is an easy way to
generate:
draw common plots from• a'line'
Pandas
: lineseries
plot
• 'bar' : vertical bar
plot
pd.Series.plot(kind='line'• 'barh' : horizontal bar
figsize=None, plot
• 'hist' : histogram
Pandas Plots title=None, • 'box' : boxplot
grid=None, • 'kde' : Kernel Density
legend=False, Estimation
plot
xticks=None, yticks=None,
• 'density' : same
as 'kde'
xlim=None, ylim=None,
• 'area' : area plot
colormap=None)
• 'pie' : pie plot
pandas.plot:
pandas.pydata.org/pandas-
docs/stable/generated/pandas.S
eries.plot.html

60

25
 13/10/2022

▪ The Pandas plot function is an easy way to

draw common plots from a Pandas series

populations.plot(kind = "bar")

Pandas Plots

61

▪ The Pandas plot function is an easy way to

draw common plots from a Pandas series
A tuple
giving the
pd.Series.plot(kind='line' width of the
figsize=None, plot in
inches
Pandas Plots title=None, (width,
grid=None, height)
legend=False,
xticks=None, yticks=None,
xlim=None, ylim=None,
colormap=None)
pandas.plot:
pandas.pydata.org/pandas-
docs/stable/generated/pandas.S
eries.plot.html

62

26
 13/10/2022

▪ The Pandas plot function is an easy way to

draw common plots from a Pandas series

pd.Series.plot(kind='line'
figsize=None, The chart
Pandas Plots title=None, title text
grid=None,
legend=False,
xticks=None, yticks=None,
xlim=None, ylim=None,
colormap=None)
pandas.plot:
pandas.pydata.org/pandas-
docs/stable/generated/pandas.S
eries.plot.html

63

▪ The Pandas plot function is an easy way to

draw common plots from a Pandas series

pd.Series.plot(kind='line'
figsize=None,
Pandas Plots title=None, Should grid
lines be shown -
grid=None, True or None
legend=False,
xticks=None, yticks=None,
xlim=None, ylim=None,
colormap=None)
pandas.plot:
pandas.pydata.org/pandas-
docs/stable/generated/pandas.S
eries.plot.html

64

27
 13/10/2022

▪ The Pandas plot function is an easy way to

draw common plots from a Pandas series

pd.Series.plot(kind='line'
figsize=None,
Pandas Plots title=None,
Should a
grid=None, legend be
legend=False, shown (True
or False)
xticks=None, yticks=None,
xlim=None, ylim=None,
colormap=None)
pandas.plot:
pandas.pydata.org/pandas-
docs/stable/generated/pandas.S
eries.plot.html

65

▪ The Pandas plot function is an easy way to

draw common plots from a Pandas series

pd.Series.plot(kind='line'
figsize=None,
A sequence
Pandas Plots title=None,
of values to
grid=None, use for the
legend=False, x or y axis
tick marks
xticks=None, yticks=None,
xlim=None, ylim=None,
colormap=None)
pandas.plot:
pandas.pydata.org/pandas-
docs/stable/generated/pandas.S
eries.plot.html

66

28
 13/10/2022

▪ The Pandas plot function is an easy way to

draw common plots from a Pandas series

pd.Series.plot(kind='line'
figsize=None,
Pandas Plots title=None,
grid=None,
legend=False,
xticks=None, yticks=None,
xlim=None, ylim=None,
A tuple giving
colormap=None)
the x or y axis
pandas.plot:
pandas.pydata.org/pandas- display limits
docs/stable/generated/pandas.S
eries.plot.html

67

▪ The Pandas plot function is an easy way to

The name of a Matplotlib
draw common plots from a Pandas
colormap objectseries
to use to
colour the chart. Examples
include:
pd.Series.plot(kind='line'
• Blues
figsize=None,• Greens
Pandas Plots title=None, • plasma
• summer
grid=None,
• seismic
legend=False,• Dark2
Accent
xticks=None,• yticks=None,
• terrain
xlim=None, More
ylim=None,
info at:
matplotlib.org/users/colorm
colormap=None)
pandas.plot: aps.html
pandas.pydata.org/pandas-
docs/stable/generated/pandas.S
eries.plot.html

68

29
 13/10/2022

Pandas Plots

1. Data
Visualisations
in Python

69

Now You 1. Simple Data

Try … Visualisations
In Python

70

30
 13/10/2022

▪ The Pandas plot function is a wrapper of specific

functions for each type of plot
- pandas.Series.plot.area
- pandas.Series.plot.bar
- pandas.Series.plot.barh
- pandas.Series.plot.box
pandas.Series.plot.density
Pandas Plots -
- pandas.Series.plot.hist
- pandas.Series.plot.kde
- pandas.Series.plot.line
- pandas.Series.plot.pie
▪ These functions give some extra options for more
control over chart appearance so are worth
exploring
Pandas Plotting:
https://pandas.pydata.org/pandas-
docs/stable/reference/plotting.html

71

Tableau Demo

▪ Now we will jump

into Python and see
how create simple
visualisations

72

31
 13/10/2022

Understanding
Perception Theory and
Visual Encodings

73

▪ In this section we will cover

- How we can use understanding of our
visual perception to create more
Section effective visualization
Overview - Understanding visual encodings

74

32
 13/10/2022

Let’s Start
with
some Hans
Magic

75

Let’s Look at
the
Characteristics
of the Chart
Position

76

33
 13/10/2022

Let’s Look at
the
Characteristics
of the Chart
Colour

77

Let’s Look at
the
Characteristics
of the Chart
Size

78

34
 13/10/2022

Let’s Look at
the
Characteristics
of the Chart Animation

79

Why did
Hans choose Lets see if we can uncover why these
these specific specific data encodings were chosen
encodings? by Hans?

80

35
 13/10/2022

SEEING IS BELIEVING?

81

Stare at the cross in the middle of the image

and think about circles
82

36
 13/10/2022

Can you see

an old lady or
a young lady?

83

For more great illusion examples take a look at: http://web.mit.edu/persci/gaz/

84

37
 13/10/2022

Which Tree Is Biggest On This Slide?

A B C D

85

This looks
better

86

38
 13/10/2022

Perception

87

▪ Our five senses (vision, hearing, taste,

smell & touch) provide us with the raw
stimulus information with which to be
aware of and understand the world
- Our sense organs are passive and
Perception biologically understandable

88

39
 13/10/2022

▪ Perception is the psychological process of

actively selecting and organising stimulus
information detected by our senses so as to
create conscious awareness
- Perception is an active process that is only
somewhat psychologically
Perception understandable
- A small subset of stimulus information is
selected and organised to create
conscious perception
- Psychophysics is the study of this process
and understanding a little about it helps us
create better visualisations

89

VISUAL ATTENTION

90

40
 13/10/2022

Visual Attention
▪ Visual attention refers to the various mechanisms that
help determine which regions of an image are selected
for detailed analysis by our vision system
▪ At any instant detailed information of shape and colour
is only possible in a small portion of our visual field -
about the size of your thumbnail viewed at arm’s length

a j g n t b f y d k e i f h t a j g n t b f y d k e i f h t

d t n j f o s e w b h t w u i d t n j f o s e w b h t w u i

y b a s g r c h y i j n g f s y b a s g r c h y i j n g f s

y v f d s p l b n h g h n j n y v f d s p l b n h g h n j n

Healey, Christopher and Enns, James, Attention and Visual Memory in Visualization and Computer Graphics, IEEE Transactions on Visualization and Computer Graphics,
vol 18 (7), pp 1170--1188, 2012
www.csc.ncsu.edu/faculty/healey/download/tvcg.11.pdf

91

▪ To overcome this limitation our eyes

move rapidly alternating between:
- Fixations: during which detailed
information from a small region is
The Fixation- visible
Saccade - Saccades: a brief period of blindness
Cycle during which eyes flick to a new
location
▪ The fixation-saccade cycle repeats 3–4
times each second
▪ What does this mean for perception?

92

41
 13/10/2022

▪ There are a limited set of visual features

that are detected by low-level, fast-acting
visual processes within the period of a
single fixation
Pre-attentive
Properties
▪ These are known as pre-attentive
properties

▪ These are some of the most effective

a j g n t b f y d k e i f h t
properties we can use in visualisations as
d t n j f o s e w b h t w u i
y b a s g r c h y i j n g f s they allow elements to pop out
y v f d s p l b n h g h n j n

93

- length - stereoscopic depth

- size - position

Common,
Useful Pre-
- shape - colour (hue)
Attentive
Properties

Reference: Healey, Christopher and

- direction of motion - colour (luminance)
Enns, James, Attention and Visual
Memory in Visualization and
Computer Graphics, IEEE Transactions
on Visualization and Computer
Graphics, vol 18 (7), pp 1170--1188,
2012
www.csc.ncsu.edu/faculty/healey/dow
nload/tvcg.11.pdf

94

42
 13/10/2022

VISUAL ENCODINGS

95

Fundamentally data visualisation is about mapping data

dimensions to visual encodings

96

43
 13/10/2022

Pre-attentive
Encoding Example Chart Types
property
Scatter Plot, Geo.
Position Yes
Map
Length Yes Bar Chart

Angle No Pie, Donut Plot

Area Yes Tree Map

Shape Yes Scatter Plot

Colour Yes Heat Map

Density Yes Heat Map

Pattern Yes Line Chart

97

Understand Visual Perception-

Choosing visual encodings based on
Top-Tip pre-attentive properties will make
your visualizations more effective

98

44
 13/10/2022

Data Visualisations Zoo

99

▪ We can group different visualisations into

the following general types:
The - Distributions
Visualistion
- Trends over time
Zoo
- Relationships
- Maps
- Data tables

"A Tour through the Visualization

Zoo", Jeffrey Heer, Michael Bostock &
Vadim Ogievetsky, Queue, Volume 8
Issue 5, 2010
http://queue.acm.org/detail.cfm?id=1805128

100

45
 13/10/2022

The Data
Visualisation
Zoo

101

Geography
Composition
The Data Highlight
Trends in a
Visualisation Tabular
Zoo Structure

Composition
Trend over
Time
Precision is
Important

102

46
 13/10/2022

You have been asked to show Sales Numbers for

different countries for the month of May in a report
and you want to create a chart for this - how do you
choose the right chart?

An Example

103

Total Sales by Country Total Sales By Country

A B C D E A B C D E
€700,000 €2,500,000
€600,000
€2,000,000
€500,000
€400,000 €1,500,000
€300,000 €1,000,000
€200,000
€500,000
€100,000
€0 €0
France Ireland United Kingdom France Ireland United Kingdom

Total Sales by Country Total Sales by Country

E D C B A A B C D E
100%

80%
United Kingdom
60%

Ireland 40%

20%

France 0%
France Ireland United Kingdom
€0 €100,000 €200,000 €300,000 €400,000 €500,000 €600,000 €700,000

104

47
 13/10/2022

DATA VISUALIZATION
ZOO

105

▪ In designing effective data visualisations, we must

take into consideration four main questions:
1. What is the key question or questions a specific
Four Main visualisation, chart or graph will answer?
Questions to 2. What data features are required to answers
Consider these questions?

3. Who is going to be making decisions based on

the visualisation?

4. How will the visualisation be presented, paper

based report, presentation or interactive
dashboard?

106

48
 13/10/2022

Tableau Demo

▪ Now we are going to

step through the data
visualization zoo in
Tableau
▪ The explanation for
which chart type to
use is detailed below

107

Making Visualisations
More Effective

108

49
 13/10/2022

▪ This Section will cover the following topics

- Colour theory
- Data Ink Ratio

Section
Outline

109

COLOUR THEORY

110

50
 13/10/2022

Colours should be used with care and

considerations, items that we need to consider
are:
- Colour should always have a clear purpose in
our dashboards
Considering
Colour - Colour does not have a natural ordering
- Consider colour blindness
- Consider the overall colour palette

111

A Note on
Colour

Wexler, Steve, Jeffrey Shaffer, and

Andy Cotgreave. The big book of
dashboards: visualizing your data
using real-world business
scenarios. John Wiley & Sons,
2017.
https://www.bigbookofdashboard
s.com/

112

51
 13/10/2022

▪ There are a lot of colours in the world, but

we cannot tell them apart if too similar
▪ The standard advice for using colour to
encode categories is to limit your
Distinct
selection to ideally about 6 - hopefully
Colours
no more than 10
▪ There are a range of defined
colour palettes that work well

113

115

52
 13/10/2022

117

▪ Colour blindness is an inability, or

decreased ability, to perceive colour
differences under normal lighting conditions
- The most common type is red-green colour
blindness
Colour
Blindness - Complete colour blindness is extremely rare
- Usually caused be a lack of development of
cones on the retina in the eye - these are
responsible for perceiving colour
- Colour blindness occurs in about 10% of males
and less than 1% of females

118

53
 13/10/2022

Colour
Spectrums

Maureen Stone, “What about “color

blindness?””
http://www.stonesc.com/wordpress/201
0/05/what-about-color-blindness/

119

Colour Blindness
▪ There are a few simple things we
can do to address colour blindness:
- Try to use monochromatic
scales when using colour
with continuous data
- Use redundant encodings
whenever colour is used to
present categories - for example
use shape and colour
- There are colour blind safe
palettes that can be used http://colorlab.wickline.org/colorblind/colorlab/

120

54
 13/10/2022

Use of
Redundant
Encodings

121

Use of
Redundant
Encodings

http://www.juiceanalytics.com/

122

55
 13/10/2022

Creating
Colour
Palettes with
Coolors
Online Tool

www.coolors.co

123

Think carefully about colour, choose

your palletes with care and only apply
Top-Tip colour when it has a specific purpose

124

56
 13/10/2022

DATA INK RATIO

125

Edward Tufte

126

57
 13/10/2022

▪ The data-ink ratio is a concept introduced by

Edward Tufte
- The data-ink ratio can be used to measure how
cluttered a visualisation is
Data – Ink - The data-ink ratio is a measure of how much of
the ink used in a visualisation is representing
Ratio
the data
- Tufte refers to data-ink as “the non-erasable ink
(or pixels) used for the presentation of data”

Ref: The Visual Display of

Quantitative Data, Edward
Tufte

127

▪ The data-ink ratio is calculated as follows:

Data − Ink
Data−ink ratio =
Total ink used to print the graphic
Data – Ink
Ratio
▪ Another way of interpreting the data-ink ratio
is the proportion of a visualisation’s ink
assigned to the non-redundant display of
data-information

Ref: The Visual Display of

Quantitative Data, Edward
Tufte

128

58
 13/10/2022

Votes by Political Party 2014 Local Election

479 K
Votes by Political Party
428 K
2014 Local Election 404 K
600,000
500,000
400,000 Independents 256 K
300,000
Fine Gael
200,000
100,000 Fine Fail
- Sinn Fein 122 K
Labour

Others Fianna Fail Fine Gael Sinn Fein Labour

We have improved the data-ink ratio and in

doing so have vastly improved our visualisation

139

Show All The Data And Only The

Top-Tip
Data

140

59
 13/10/2022

Building Dashboards
that People will Love to
Use

141

Sample
Dashboards

Google Analytics:
analytics.google.com

142

60
 13/10/2022

Sample
Dashboards

https://ig.ft.com/sites/numbe
rs/economies/us/

143

▪ This Section will cover the following topics

- Defining the purpose of a dashboard
- Designing a dashboard
- Making it happen in Tableau
Section
Outline

144

61
 13/10/2022

▪ The three key questions in defining the

purpose of your dashboard are
- Who are you trying to impress?
 Know your audience
Three Key - Why bother?
Questions  Know the value your dashboard will add to your
organisation
- How will you impress your audience?
 What type of dashboard will be created

145

• What Decisions do they make?

Role
• What Questions do they need answered?

Work • How will they be reviewing the data?

Know Your Process • Currently how do they make decisions?
Audience

Technical • Are they data savvy?

Ability • How good are their data exploration skills?

Business • Are they familiar with terminology?

Prowess • Are they familiar with the company KPIs?

146

62
 13/10/2022

• Multiple Audiences - information refers to

an entire organisation
Audience • Single Audience - information refers to a
What Type Of specific department, product line, customer
Dashboard segment
Are You
Creating?
• Strategic - These dashboards display high-
level, company wide, performance driven
Business
• Operational - These dashboards display
Role
data that facilitate the operational side of a
business.

147

• Historical – tracking historical trends

Time
• Snap shot – point-in-time view of the data
Horizon
What Type Of • Real time – displaying the current situation
Dashboard
Are You
Creating?
• None – The consumer can only view the
report
Interactivity
• Drill-able - The consumer has the ability to
slice and dice the data

148

63
 13/10/2022

Actionable

Choosing The
Perfect KPI Accessible
credible Common
data Interpretation

Transparent simple
calculation

149

▪ Not all KPIs are equal in terms of importance

- KPIs should be ordered by importance and the
important KPIs should be given prominence on
the dashboard
Considering ▪ For each KPI think about the best way to
KPIs visualise it:
- Distributions
- Trends over time
- Relationships
- Maps
- Data tables

150

64
 13/10/2022

▪ Before embarking on the creation of a

dashboard or report make sure you can
Defining The answer the following questions:
Purpose Of 1. Who is the audience for the dashboard and
what are their needs?
Your
Dashboard 2. How will the dashboard inform better decision
making in my organisation?
3. What type of dashboard will best suit the task?
4. What is the theme of my dashboard?
5. What KPIs will I include in my dashboard, how
will they be visualised and their importance?

151

DASHBOARD
STRUCTURE

152

65
 13/10/2022

▪ Now that the purpose of the dashboard has

been defined in detail the next step is to
design the structure of the dashboard
▪ Consider the following three areas in
designing the structure of the dashboard:
Introduction - Form: In what format is the dashboard
delivered?
- Structure: How is the dashboard going to be
laid out?
- Functionality: How will the users interact with
the information provided?
Juice Analytics "A Guide to Creating
Dashboards People Love to Use”
www.juiceanalytics.com/s/Guide_to_
Dashboard_Design.pdf

153

▪ Number the boxes from 1 to 9 to reflect

where a user is most likely to look.

Layout &
Attention

154

66
 13/10/2022

▪ Number the boxes from 1 to 9 to reflect

where a user is most likely to look.

Layout &
Attention 1 4 6

3 2 8

5 7 9

155

Consider the Flow

156

67
 13/10/2022

Consider the Flow

157

▪ There are a number of functional features

that can make your dashboard more useful
for the user
▪ Here is a list of some basic features:
- Drill Down
Functionality - Customise
- Filter
- Comparison
- Alerts
- Export or Print

158

68
 13/10/2022

▪ In designing your dashboard keep the

following in mind
- What form will the dashboard take?
- How will the dashboard be laid out?
Structure - What functionality will be added to make the
Overview dashboard more interactive?

159

THE VISUAL ANALYTICS

PROCESS

160

69
 13/10/2022

The Visual Analytics Process

▪ We now want to look at a more specific process for creating
Visual Analytics Solutions
- We have adapted a process outlined by
Ben Fry in: “Visualizing Data: Exploring
and Explaining Data with the Processing
Environment”

“Visualizing Data: Exploring and Explaining Data with the Processing Environment”, B. Fry, O'Reilly Media, 2007
http://shop.oreilly.com/product/9780596514556.do

161

▪ Considerations for designing dashboards:

Analytics - Every solution will be made up of a
Solutions To number of smaller components
Specific
- Each of these small components should
Questions
map to a specific question
- The visual analytics process is driven by
questions

162

70
 13/10/2022

The Visual Analytics Process

Define Acquire
Represent Explain Refine Present
Questions Data

163

The Visual Analytics Process

Define Acquire
Represent Explain Refine Present
Questions Data

Based on the business

problem outline a set
of questions that will
form an analytics
solution

164

71
 13/10/2022

The Visual Analytics Process

Define Acquire
Represent Explain Refine Present
Questions Data

Obtain the data

required to
answer these
analytics
questions

165

The Visual Analytics Process

Define Acquire
Represent Explain Refine Present
Questions Data

Choose a basic
set of visual
encodings, such
as a bar graph,
line graph, tree
etc.

166

72
 13/10/2022

The Visual Analytics Process

Define Acquire
Represent Explain Refine Present
Questions Data

Explain the
patterns visualised
to answer the
analytics
questions

167

The Visual Analytics Process

Define Acquire
Represent Explain Refine Present
Questions Data

Refine the
visualisations created
to better understand
the data or answer
different questions

168

73
 13/10/2022

The Visual Analytics Process

Define Acquire
Represent Explain Refine Present
Questions Data

Present
visualisations to
business
stakeholders

169

The Visual Analytics Process

Define Acquire Represent Refine

Explain Present
Questions Data

“Visualizing Data: Exploring and Explaining Data with the Processing Environment”, B. Fry, O'Reilly Media, 2007
http://shop.oreilly.com/product/9780596514556.do

170

74
 13/10/2022

Following the Visual Analytics

Process, focusing on Questions,
Top-Tip
helps us develop effective
visualization

171

▪ The key points made in this section were:

- The three key questions in defining the purpose
of your dashboard are
 Who are you trying to impress?
Why bother?
Section 
How will you impress your audience?
Takeaways


- In designing your dashboard keep the following

in mind
 What form will the dashboard take?
 How will the dashboard be laid out?
 What functionality will be added to make the
dashboard more interactive?

172

75
 13/10/2022

Tableau Demo

▪ Lets look at how we

can create
dashboards in Python
and Tableau

173

Workshop

▪ Now you can try to

create a dashboard
with our banking
data.

174

76
 13/10/2022

Thank You

175

References
1. Visualizing Data 3. Beautiful 4. The Big Book of
Ben Fry Evidence Dashboards
Edward Tufte Wexler, Shaffer &
Cotgreve

2. Semiology of 4. Beautiful
Graphics Visualisation
Jacques Bertin Steele & Illinsky

176

77
 13/10/2022

▪ Thank you for attending this course

▪ If you have any questions about how to apply
what you have learned, or need clarification
on any topics, please contact us at:
Thank You - Email : ecb-learners@krisolis.ie
- Web : www.krisolis.ie
- Twitter : @KrisolisHQ

177

78