2018-07-28_viz_talk

Transcript

1. Data Viz Language, Perception, and in Practice 2018-07-28 leoluyi@iii Slides https://leoluyi.pse.is/iiiviz © leoluyi, 2018 1
2. We'll discuss about... 4 What's a good viz? 4 A brief history of data viz. 4 Real world implementation 4 Get better with a simple framework © leoluyi, 2018 2
3. 橕ෝ౯ 4 㸎瓽 Leo Lu 4 Management + Psychology 4 CRM for Financial Service 4 Build data products 4 ETL 4 Models 4 Text mining 4 Viz 4 ... © leoluyi, 2018 3
4. It's all about data Charts, graphs, maps, diagrams, ... © leoluyi, 2018 4
5. Speak the language © leoluyi, 2018 5
6. Trends driving the need of visual thinking 1. Massive increase of visualization 2. Data 3. Everybody’s doing it © leoluyi, 2018 6
7. Past and Present © leoluyi, 2018 7
8. Brief history of dataviz 4 (1960) Bertin: Visual variables 4 (1970) Tukey: Exploratory Viz 4 (1980) Tufte: Design principles for information (Chartjunk) 4 (1980) Cleveland & McGill: Graphic perceptions 4 (2000) Computer-driven, Design-driven 4 (2010) Easy-to-use apps 4 (Now) Interactivity, dynamic update © leoluyi, 2018 8
9. Perceptions in psychology © leoluyi, 2018 9
10. Perceptual tasks1 1 In several experiments, Cleveland and McGill (1984) identified “elementary perceptual tasks”: the most basic tasks they believe viewers perform when evaluating a visualization. © leoluyi, 2018 10
11. Color © leoluyi, 2018 11
12. Color over shape © leoluyi, 2018 12
13. Reading texts © leoluyi, 2018 13
14. Reading charts - We don’t go in order © leoluyi, 2018 14
15. And more other else 4 People see first what stands out. 4 People see only a few things at once. 4 People seek meaning and make connections. 4 People rely on conventions and metaphors. 4 ... © leoluyi, 2018 15
16. A good chart? © leoluyi, 2018 16
17. © leoluyi, 2018 17
18. Viz works in the real world ܉虋苭玕瞤犋螂聲樿ጱӞݙ扖 © leoluyi, 2018 18
19. Data Management Will you change your data? (eg. changing one or all values or adding rows or columns) © leoluyi, 2018 19
20. ETL "You can't do viz without those data skills!" ➔ Autonomy? DataOps © leoluyi, 2018 20
21. ፓጱ究ਧಋྦྷ 4 Analysis vs. Presentation 4 R, Python <-> D3.js, Illustrator 4 Chart typologies vs. innovative outside-of-the-box charts 4 Excel <-> D3.js 4 Interactivity vs. static 4 PowerBI <-> PowerPoint © leoluyi, 2018 21
22. There Are No Perfect Tools Just Good Tools for People with Certain Mindsets © leoluyi, 2018 22
23. Get better with a simple framework © leoluyi, 2018 23
24. Viz Quadrant 1. Conceptual (Idea-Driven) vs. Data-Driven 4 Do I have ideas or data? 2. Declarative vs. Exploratory 4 Show what vs. Show why Berinato (2016). Good Charts 24
25. Conceptual © leoluyi, 2018 25
26. Data-Driven © leoluyi, 2018 26
27. Berinato (2016). Good Charts 27
28. Other Tips © leoluyi, 2018 28
29. Ask and Listen 4 Talk to your stakeholders 4 Keep records of words, phrases, and statements © leoluyi, 2018 29
30. Prototypes © leoluyi, 2018 30
31. Recap - Make dataviz smoother 4 Sharpen your viz thinking 4 Principles 4 Process 4 Data management 4 Availability 4 Cooperate with data engineer, data scientist 4 Familiar with your tools © leoluyi, 2018 31
32. ൉㺔膏Ի窕 㸎瓽 leoluyi@github https://leoluyi.github.io © leoluyi, 2018 32