Loading presentation...

Present Remotely

Send the link below via email or IM


Present to your audience

Start remote presentation

  • Invited audience members will follow you as you navigate and present
  • People invited to a presentation do not need a Prezi account
  • This link expires 10 minutes after you close the presentation
  • A maximum of 30 users can follow your presentation
  • Learn more about this feature in our knowledge base article

Do you really want to delete this prezi?

Neither you, nor the coeditors you shared it with will be able to recover it again.


Intro to Data Visualisation for Tableau users

My introduction to data visualisation workshop which I run at work

Paul chapman

on 19 July 2016

Comments (0)

Please log in to add your comment.

Report abuse

Transcript of Intro to Data Visualisation for Tableau users

Transforming information into insight
Introduction to Data Visualisation
Data Visualisation
Think first
Visualise second

Information v Insight
Known for his conversational maxims, describing the characteristics of polite conversation, each of which can be applied equally well to communicate quantitative information in the workplace via tables and graphs
Paul Grice a 20th Century Philosopher
This is the kind of graph that software products, including Excel, encourage us to create. They give us an infinite selection of poorly designed graphs from which to choose. What we really need, however, is a small selection of graphs that really work.

Using this graph, try to see the pattern of change across the months in actual values. Try to determine one of the actual values. Try to compare actual to budget over time.

Let's transform this graph into one that communicates...
Get to the point
3 simple rules:

1) Clarify the purpose
Why are you sharing this insight?

2) Simplify the message
Remove all the noise

3) Drive the outcome
Insight should drive action

Information requires interpretation whereas insight requires action.

The difference between information and insight is often the way it is presented to and perceived by the audience - hence the importance of data visualisation.

Information is not the same as insight.

Information tells you that the chicken crossed the road
whereas insight tells you why the chicken crossed the road
Insight often points to other lines of enquiry like...

Why did the duck cross the road?
Bore or
Data visualisation can either:
Facilitate action
Benefits of getting your visualisation right:
Focuses discussion around the important points of your message
Makes data driven decision making easier
Stimulates stakeholders to respond
“The ability to design effective visual displays of data is not intuitive; it requires a set of visual design skills”.
(Stephen Few)
An effective visualisation of your data helps clearly identify and draw attention to issues which require a decision; thus driving the organisation forward. It is the call to action which should shape your choices when creating the most suitable data visualisation.
William Playfair was one of the first people to tell a story using a graph, making the case in 1786 against England's policy of financing colonial wars through national debt
Graphs are now commonplace, 229 years later, partly due to the arrival of the PC, fully integrated into the fabric of modern communication.

However, this makes us feel like experts in graph design which couldn't be further from the truth - especially for news organisations.
Graphs today
In case you think the Onion's spoof newscast is an exaggeration
Today, vastly more misinformation is disseminated unintentionally because people do not know how to use charts to communicate what they intend to.
One Common problem with the display of quantitative information is that people often choose the wrong medium of display - a graph when a table would work better and vice versa.
Pie Charts
The purpose of this graph is to display how Company G is doing in relation to its competitors.
Is the message clear?
You find that when someone creates a graph that appears inadequate, they try to fix it with sex appeal
By adding 3D and lighting effects the chart becomes harder to read
Design Choice matters when telling a story
Remove thick borders
Have a precise scale
Single labels for currency
Orient the legend in same order
Focus colour on the actuals
Bar graphs should ALWAYS start at zero
Reset scale when using lines
One chart alone may not be the answer
Do not try and fit onto one graph

Tables and graphs help us ........
Four Categories
Make your contribution to the conversation informative
Do not make it more informative than necessary
Do not say what you believe is false
Do not say that for which you lack adequate evidence
Be relevant to the current topic of conversation
Avoid obscurity of expression
Avoid ambiguity
Be brief
Be orderly
Determine the medium that tells the story best
Design the components to tell the story clearly
Tables work best when:
used to look up individual values
used to compare individual values
Data must be precise
You must include multiple units of measure
You want to show both details and summaries
Variance of size and relevance are not an issue
Graphs work best when:
used to feature patterns, trends, exceptions and entire series of values at once
Compare sets of data
Show the minimums and maximums quickly
Any data we present is made up of two parts -
Quantitative Values (Measures in Tableau)
Categorical Labels (dimensions in Tableau)

Quantitative values are the measures of something related to the business -
Sales, capacity, RPS, contribution
Categories break the measures down into meaningful groups and provide meaningful labeling
Scales on a graph are either quantitative or categorical, three of which are common in graphs -
Nominal: Individual items along the scale differ in name only and do not have a particular order and represent no quantitative values
Sales; Operations; Finance
Ordinal: The items along the scale have an order of rank but do not represent quantitative values
1st; 2nd; 3rd; 4th
Interval: The items along the scale have an intrinsic order which does correspond to quantitative values
(0-99, 100-199, 200-299, January, February, March)
Things to avoid -
Pie Charts
Use Treemaps for vast data
Keep Backgrounds Clean
Colour Can be distracting
Avoid Shapes
Remove tick marks
Eight common relationships in Graphs -
Time Series - Changes through Time
Ranking - Relative magnitude in order
Part to whole - 100% sum data
Deviation - Variance to plan or to each other
Distribution - Order split across range
Correlation - Relationship between two sets of values
Geospatial - Geographical location of each value (can use a pie!)
Nominal Comparison - Values compared across a nominal scale
Time Series
Part to Whole
Nominal Comparison
Data - Ink Ratio
Tables and graphs are made up of two types of ink
Ratio of ink used to display data should be high
with all other ink reduced to a minimum
Our eyes and visual cortex are massive pattern seekers
Massive data processor
We can see some patterns when presented one way
which can be invisible to us if presented differently
Eyes do not register colour but the differences in colour between something and what is nearby
Context also effects what we see
This shows how the background colour change
of an image effects our perception of colour
We perceived the attributes pre-attentively which means it takes place in advance of thought which is extremely fast
Some visual attributes are easier to see than other
When presenting data visually, these are the most powerful and useful to us
Our eyes allow up to interpret the individual attributes of a picture we are seeing
It can be easy to not see what is in front of you
because you are looking for something you already believe to be there
The brain focuses all of its efforts on a single task
Lets try that again ..........
Use least visible means to support the function of the non data ink
Increase white space
Numbers should always be right aligned
Decimals should be consistant
Text should always be left aligned
When highlighting sections of data in a a table use
a Border to create an enclosure
Fill colour which should be a contrasting hue
BOLD text for greater data width
Coloured text should be a contrasting hue
2D xy graphs
Four different objects can be used to visually encode data within graphs
Has no dimensions as it marks a point but has no size or width
Can be thought of as a point which has been extended in a direction
A line to which an additional dimensino of width has been added
Like a bar except both ends are used to mark a value
Six variations of these four work
well in XY graphs
Good at comparing the magnitude of different objects.
Able to easily emphasise individual discreet values
Must always begin at a value of zero
Otherwise length does not correspond accurately to their quantitative value
Good at comparing the shape of change from one value to the next through time using an interval scale
Should only be used to display continuous numbers along an interval scale
Good at encouraging us to notice clusters of values.
Useful in correlation relationships and do not require a scale
Good for showing two or three relative values at the same time
Represent the length of a full spread of the values, from highest to lowest
A mark is used to divide the box into sections
A box is a combination of both points and bars (Commonly known as a box plot)
We have the rule of not encoding data as a 2D area
The brain cannot accurately compare the size of areas
How many times bigger?
Which visual objects does the best for each type of quantitative relationship?
Nominal Comparison
Bars should be used because they emphasis the independent nature between each nominal
Time series
Lines do a greater job in showing the flow of values across time
Use bars if you want your message to emphasise individual values
Bars of points can be used - first sort your data to put the data in a ranked order to make it easier to compare
Part to Whole
Hopefully you know why not to use a pie chart!
Use bars, ensuring that it is clear on the graph that all the parts add up to 100%
Can use stacked, but only the bottom stack can be accurately measured
The use of a reference line should be used to make it clear that the main point of the graph is to display how one of more measures deviate from a point of reference
Histrograms and frequency polygons can be used
When you only have a few data points strip plots can be used
Box plots are usually the best way of comparing multiple distributions
Scatterplots using points and trend lines are the best way
Simplist way is to display data points at each location
Vary their size through bubbles to show differences
Colour can also add a second set of measures
You can also display many graphs within a single eye span
This sounds unmanageable for the brain, but can be easy to understand ....
Related graphs can be arranged in a row, column or a matrix known as
They must vary nothing but a single variable and arrange the graphs in a logical sequence,
usually based on the ranked values
The definitive Story telling moment using Graphs
Hans Rosling presented a compelling quantitative story at the TED conference for the first time in 2006
This compares life expectancy against GDP since 1800
Which square is darker?
A or B?
The answer is ......
they are BOTH the same
Are the rectangles different colours?
How many times does the number 5 appear above?

You have 3 seconds to work out the answer
Now they are highlighted you get the answer in under a second
Did you get it?
Key takeaways
Full transcript