Delhi.bootcamp dataviz
•Download as PPTX, PDF•
1 like•2,718 views
Dataviz presentation for http://delhi.dbootcamp.org/
More Related Content
Similar to Delhi.bootcamp dataviz
Similar to Delhi.bootcamp dataviz (20)
Recently uploaded
Recently uploaded (20)
Delhi.bootcamp dataviz
- 1. “I can’t see the wood for the trees” (7 things to help make data-viz clearer)
- 2. Humans are intensely visual creatures
- 3. Location Alphabet Time Category Hierarchy Richard Saul Wurman
- 4. Data visualisation is about exposing patterns
- 5. Correlation != causation Per capita consumption of mozzarella cheese (US) correlates with Civil engineering doctorates awarded (US) Spurious Correlations: http://tylervigen.com/view_correlation?id=3890
- 7. What does your audience want? User Stories As a <someone>… I want <a feature>… So that <a benefit>…
- 8. User Stories As an Hispanic Mother I want to see how unemployment disproportionately affects young women who haven’t graduated high-school, So that I can show my daughter the difference an education makes http://nyti.ms/NHqEMC
- 9. Overview first, Zoom and Filter, Details-on-demand Impact curiosity exploration personalisation Ben Schneiderman
- 10. Keep It Simple S…
- 11. Less is more http://bit.ly/1nC9JWO
- 12. Size matters 3 2 1 0 Xx Xx πx² / 4
- 13. Don’t use two dimensions when one will do
- 14. Worse still… Don’t use 3D What are 3D Pie charts good for?
- 15. The axes of evil Changing y-axis maximum Changing ratio of graph dimensions
- 16. The axes of evil Truncated bar graphs http://bit.ly/1tcUsig
- 17. Colour Matters 1 in 12 men are colour blind http://wtfviz.net/
- 19. Make it personal I've just found out I’m in the Traditional working class group in Britain’s new class system #Whatsyourclass http://bbc.in/12acLLV Overpriced fuel? I pay £0.39 more than average to fill up. Try @BBCNewsGraphics price calculator http://bbc.in/WeKlw4 #petrolprice Which Olympics athlete are you? I'm Paloma Schmidt - find out yours with the BBC's #AthleteLikeMe #BBC2012 #London2012 http://bbc.in/NE48lq
- 20. 1. Form follows function (What are the patterns in your data) 2. What does your Audience want? (User stories: “As x, I want y, so that z”) 3. Overview first, then detail (Martini glass) 4. Keep it simple (Less is more) 5. Visualisations can misrepresent and mislead (Shapes, axes and colour) 6. Geography != maps (Maps are great – but complicated) 7. Make it personal (What’s the Hashtag?) More links: one-tab.com/page/r-QbXpgPR5KmyRdq0sYnEg
Editor's Notes
- Information is complicated and messy and often overwhelming. Even if you know your data contains useful information, sometimes there is just too much of it to make much sense of. Information that really matters is often buried amongst a lot of data that just isn’t irrelevant Of course what’s relevant to me, might not be relevant to someone else, but we’ll get to that later.
- The human brain is programmed to find patterns. Visualizing data is the fastest way to find patterns AND these patterns MIGHT lead you to a story. (OR NOT… be careful – if in doubt ask a statistician!) Visualizing data is also the fastest way to communicate it to others. Time is a valuable commodity, as journalists you want to get your story across efficiently particularly if your users are on mobile phones Also you want some thing that has visual impact and will encourage your audience to share your story (again more later) Data Visualisation is often a great way tofind a story and tell a story…
- Normally data will have had some Method of Organization associated with it OR part of your job as a data journalist will have been to find the most appropriate way to interrogate the data LATCH: Richard Saul Wurman Alphabet Forced organization rather than natural. Time Easy to understand, easy to draw comparisons and conclusions. a narrative Category Well reinforced by color & placement. Grouped by similar importance – a value judgment. Hierarchy Assign value or weight to the information; usually on a scale largest => smallest $$$$ => $
- The Latch method of organisation will give you some ideas about ways to sort your information to expose patterns. By visualising those patterns you make the data easier to understand. But you will only make the data easier to understand if the visual shapes you use come from the data and NOT the other way round. FORM FOLLOWS FUNCTION Dataviz is used to communicate a message that is contained in the shape of the data Is used to reveal relationship among many values Note about most examples being from the BBC (sorry) but NYT = kings of Dataviz Jobless like you 2009 http://www.nytimes.com/interactive/2009/11/06/business/economy/unemployment-lines.html?_r=0
- Just because two things look like they might be related – it doesn’t mean they are. Spurious Correlations http://tylervigen.com/
- This is one of may favourite pointless pieces of dataviz. Guess which states in the US have the most shark attacks. Sad thing is – it probably took a bit of work to build this and basically ended up getting ridiculed on twitter for days. Of course the data about Shark attacks in the US almost certainly DOES have some stories in it… Up or down, what age are they - even a geographical story about specific beaches or something.
- I’ve seem some people quoting Ben S… I think its similar to model I come first heard in a presentation by a designer at the Guardian The Martini Glass.. Every death on every road: http://www.bbc.com/news/uk-15975562 I saw “Amanda Cox” quoted as saying … What we don’t do is JUST say “here’s the data have play”. Your journalism isn’t working very hard for the user if they then have to do too much of their own investigation. That said – Making is story personal is something we’ve found REALLY helps the storytelling and obviously does require the user to tailor the data to their situation.
- Basically, if you see a 3d pie chart you know the person who did it doesn’t understand dataviz
- Of all the ways to mislead with statistics, this is probably the most frequent.
- http://seeingcomplexity.wordpress.com/2012/08/03/using-statistics-to-lie-and-why-democracy-needs-statistical-literacy/ Scary looking tax rise or almost meaningless tax rise. The most terrible examples of this are when very small differences are made to look big by
- Colour can be misleading – there are worse examples of this. Colour can look horrid https://kuler.adobe.com/ http://wtfviz.net/ http://www.lighthouse.org/accessibility/design/
- Just because you have some geographical data it doesn’t automatically mean you should use a map. Here Circles again! Can’t tell where they are Can’t see any hierarchy REMEMBER: communicate a message that is contained in the shape of the data Reveal relationship among many values The graph works a lot better – they were together (maps do look nice) There are other ways to illustrate geographical information – NHS Winter. http://www.bbc.co.uk/news/business-23234033 sometimes maps are great Corruption – successful – point out UX issue Where can I afford to live: Show personalisation - tweet Here the map is secondary to the story but important enough to appear on the page – but an ordinary map would have distorted he data. In the UK parliament there is 1 MP per constituency – so they all should be the same size. http://news.bbc.co.uk/1/hi/uk_politics/election_2010/8609989.stm
- As I already mentioned with the housing map and the hospital data, finding the stuff that’s about you really helps tell the story. It also encourages sharing. UK is largely focused on FB and Twitter – India? We try and build this into the story if possible and we’ve had some of our biggest successes in terms of traffic with this type of content. (class over 10 million uniques – athletes close to that) I’m x what are you?
- Point 6 Two of the most intelligent developers I’ve worked with were experts in this area – It’s a tough area to get really good at, but you can do some simple things quite quickly Next session