Introduction to effective data visualisation ( #datavis ) techniques, delivered to an HR Data Analytics conference hosted by Tucana, in London, April 2014. These slides have been adjusted since the original presentation, to mitigate confidentiality or privacy concerns. Data included herein, should be considered as Dummy Data. Ref's provided.
Disclaimer: Although I worked for IBM, and represented IBM at the time, subsequent opinions should not be considered to be in line with those of IBM.
Or F. “Depends on whether you’re making something I like…!”
Photograph is speaker’s own
Or F. “Depends on whether you’re making something I like…!”
Photograph is speaker’s own
Humble beginnings to the visualisation of data – but imagine the advantage a General would have when able to turn information into insight!
Charles Minard (1781-1870)
Fast forward 200 years…
http://www.senchalabs.org/philogl/PhiloGL/examples/worldFlights/
But we needn’t over-complicate matters – often, a simple chart is the right answer.
Global Human Capital Trends 2014: Engaging the 21st-century workforce, p10 - A report by Deloitte Consulting LLP and Bersin by Deloitte
But if wanting to feed your creative sides, unlock value by visualising new areas for advantage, rather than spending time and energy on gimmicks!
Social interactions around Mark T Lawrence (subset)
This simple chart, in itself, has demonstrated real power and traction with my own stakeholders.
Check against the 4 Pillars Model: Purpose > Content > Structure > Format
Green = Data = Innovation that matters
Purple = Cloud = Trust and personal responsibility in all relationships
Blue = Engagement = Dedication to every client’s success
Diagram by Mark T Lawrence, IBM
In this globalised world, there are still advantages to be sourced from local processing!
Diagram by Mark T Lawrence, IBM
Call out the differences? Colours; Font Sizes and Spacing (right viz has an extra chart and an extra headline)…
But: Does anyone notice that the number of people is different in the first histogram? Right = 69/126 (54.7%); Left = 55/100 (55%)
In short, Infographics are novel and have an aesthetic appeal, in the same way that a newspaper article might – but they may also be presenting subliminal bias, in the same way, too.
‘Attentional Blindness’ – Fast & Slow (System 1 and System 2) = Habitual Decision-making leads to predictable errors
‘Confirmation Bias’ – Looking for patterns which prove what we think we know
‘Risk Aversion’ – How a problem is framed affects the outcome
Hans Rosling, quoted by David McCandless, see slide 26 for reference
How the mind makes sense of data - Choose the right visualisation to get your message across
Diagram by Mark T Lawrence, IBM
Colin Ware (‘Information Visualization: Perception For Design’ 2nd ed. Morgan Kaufmann), quoted by Stephen Few (p75 ‘Show Me The Numbers’)
55% chose the format which is so familiar to us now;
8% chose the calculator layout (789, 456, 123, 0);
7% chose to track downwards from top-left
Diagram by Mark T Lawrence, IBM
Example Visualisations by Mark T Lawrence, IBM [Dummy Data]
Gestalt Institute: “Rule of Proximity”
Example Visualisations by Mark T Lawrence, IBM [Dummy Data]
Example Dashboards by The Economist Intelligence Unit, in partnership with Wipro (distributed publicly)
What do we like about this dashboard?What Don’t We Like?
Clear charts and contrasting makes easy definitionIs the most important chart, the first you see?
Available on webWhat are the charts measuring?
Sharing via social mediaWhere is the data? How can we verify or drill?
Nice banner; fits to one page Does it fit to one page? Where are the navigational breadcrumbs?
There is no link between Geo Split and Country Split: here I try to drill to the UK, but I’m unsure if the other charts updated. I can see Geo Split didn’t; how do I know if Retailer Split and Functional Split are showing UK only or worldwide?
Given this, do we think that the map is the most important message? In most cultures (particularly western), our brains are conditioned to follow a reading order (left-right, top-bottom), so the first we see is likely to be the top-left. It doesn’t appear to be linked to anything else, so what value is it?
In fact, what is the map telling me?
Example Dashboards by The Economist Intelligence Unit, in partnership with Wipro
What do we like about this dashboard?What don’t we like?
Idea of highlighted text helps reader see message The Treemap – inefficient medium for analysis
Ability to choose filter The highlighted text doesn’t relate to any charts and doesn’t update when a selection is made
Not too clutteredThe differences in bar widths between charts
(- Changing this enables slicker layout, too!)
Example Dashboards by The Economist Intelligence Unit, in partnership with Wipro
What do we like?What don’t we like?
The layout is rushed and ill-thought through
The circle chart gives inefficient message transfer
The very wide bar chart – waste of space, not comparable with other charts (more important?)
Example Dashboards by The Economist Intelligence Unit, in partnership with Wipro
WebGL Visualisation by Denny Vrandecic
IBM Whitepapers selected for sharing – all rights reserved
A search in IBM Connections shows that only 121 IBMers have a tag of Data Visualisation (out of 400,000 = 0.03% = Pacific Islands contributes only 0.03% to Greenhouse Gas Emission) – but those listed here really know their stuff!
Those named here are the personal choices of Mark T Lawrence, and are not representative of IBM. IBM is in no way responsible for the content posted by individuals outside IBM infrastructure.