The story of data visualization
0 likes176 views
The document provides an extensive overview of data visualization, tracing its history from ancient times to modern practices, and highlighting the importance of visuals in communication and learning. It includes various data visualization techniques and tools, their strengths and constraints, along with ethical considerations in presenting data. Additionally, the document discusses common pitfalls in visualization, urging accuracy and clarity to maintain audience trust.
The story of data visualization
- 1. Visuals Thestoryof An intro to data visualization
- 2. Datavisualizationisanewtrend Egyptian colored map 3,000 BC !!
- 3. Visualizationisastatisticalmethod First Chart 10th Century Astronomer Florence Nightingale - Nursery First Compound Rose Chart 1850
- 4. Animageworth1000words 0 50000 100000 150000 200000 250000 300000 350000 400000 undefined Female Male Users Joined platform since Jan 2017 Gender Count % Female 117,000 22% Male 400,000 75% Undefined 14,000 2% Male : Female 3 : 1
- 5. Human beings learned to communicate with drawing before they invented writing. Whydoweprefervisuals? Our brain can process images at speed up to 13 mms According to studies 65% of people are visual learners
- 6. Theyrevealunseenrelations First Outbreak 1817 Second Outbreak 1829 Third Outbreak 1851 Cholera pandemic
- 7. Nop, this one knows nothing John Snow 1813 – 1858
- 8. Nop, this one know nothing John Snow 1813 – 1858 Dr. John snow
- 9. John Snow Cholera Map 1850 The data visualization that revealed that water contamination is the real cause behind Cholera spread
- 12. TheQuestion? • Why do I want to do it? • What are the constraints that bound this visualization • What kind of relationship it represent?
- 13. Comparison Composition Distribution Data over time TypesofRelationships Geo based
- 14. Comparison If you are comparing numbers … Bars are your best friend 45% 30% 21% 0% 10% 20% 30% 40% 50% campaign 1 campaign 2 campaign 3 Percentage of conversion per campaign Values on top of bar Unified Color X Axis is ordered according to values
- 15. BarCharts-Colors 0% 2% 4% 6% 8% 10% 12% Linked In Indeed Wuzzuf Bayt Jobzella Tanqeeb Olx Market Growth rate To highlight information
- 16. 0% 2% 4% 6% 8% 10% 12% 14% 16% 18% Linked In Indeed Wuzzuf Bayt Jobzella Tanqeeb Olx Employer growth JS growth BarCharts-Colors To add more information/ dimension
- 17. 0% 5% 10% 15% 20% 25% Q1 Q2 Q3 Q4 Increase rate in job seekers across job categories IT Accounting Sales Tourism Education Marketing BarCharts-Colors Don’t Over do it
- 18. BarCharts-Orientation 4300 3200 2200 2100 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 سنتركول-Call Center خارجيةمبيعات وتوزيع التليفونعبرتسويق الكترونيتسويق number of job seekers joined in Feb per sub category
- 19. 4300 3200 2200 2100 0 1000 2000 3000 4000 5000 سنتر كول-Call Center وتوزيع خارجية مبيعات التليفون عبر تسويق الكتروني تسويق number of job seekers joined in Feb per sub category Long titles BarCharts-Horizontal
- 20. BarCharts-Stacked 3000 2000 800 500 1200 1400 1400 600 سنتر كول-Call Center وتوزيع خارجية مبيعات التليفون عبر تسويق الكتروني تسويق number of job seekers joined in Feb per sub category Online Registeration Call Center Compare totals & highlight composition
- 21. BarCharts-Stacked 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% سنتر كول-Call Center وتوزيع خارجية مبيعات التليفون عبر تسويق الكتروني تسويق Chart Title Online Registeration Call Center Compare Composition
- 22. BarChart-Deviation 10% -12% 15% 5% -15% -10% -5% 0% 5% 10% 15% 20% Campaign 1 Campaign 2 Campaign 3 Campaign 4 Deviation from Conversion target per campaign
- 23. Comparison-Bubbles ? ? If this bubble represent 1 then what the other bubbles equal? 1
- 24. Good in highlighting the biggest item when not comparing exact numbers Comparison-Bubbles
- 25. PieChart 70% 10% 20% Job seekers education levels Bachelor Degree Masters Degree Others Pie charts are catchy representation that is informative if the differences between numbers are significant and categories are 4 or less Values MUST sum up to 100%
- 26. Composition-Pie Bachelor Degree 40% Masters Degree 20% High School 15% Diploma 15% PHD 8% Vocational 2% Job seekers education levels Values correspond to areas, small values are not well represented
- 27. Composition-Bars 40% 20% 15% 15% 8% 2% 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% Job seekers education levels 1,500,000 JS Column height correspond to value Values are ordered descending
- 28. PieChart-Donut 70% 10% 20% Job seekers education levels Bachelor Degree Masters Degree Others 1,500,000 JS
- 29. Distribution–thecloserlook What is the average salary for a job seeker in a recruitment platform?
- 30. Distribution–thecloserlook What is the average salary for a job seeker in Forasna? Mean : 3100 LE Sum /count
- 31. Distribution–thecloserlook What is the average salary for a job seeker in Forasna? Median: 1900 Order then get the middle value
- 32. Distribution– Boxplot What is the average salary for a job seeker in Forasna? Sum /count 25% 50% 75% 99% Data can be concentrated at the end of quartile
- 33. Distribution–Histogram What is the average salary for a job seeker in Forasna? Current Salary The column represent how many times this value appeared (100 js have the salary of 2400) Bin size control the grouping of data (if size=100 then 6100 are counted under 6000)
- 34. Bin Size 100 Bin Size 500
- 35. Bin Size 100 Bin Size 1000
- 37. 0 1000 2000 3000 4000 5000 6000 Created applications per month Add trend line to highlight the over all trend Trends The slope between each 2 points represent the percentage of change
- 38. 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 Created applications per month over years Trends–Seasonal Compare similar date frames to detect seasonality 2017 2016 2015 Add legend at the end of line
- 39. Heatmaps • Sequential Data? 1 color different shades • Categorical data? Different colors • Divergent data? 2 color shades • Use Color Brewer to help with palettes.
- 40. 0% 2% 4% 6% 8% 10% 12% 14% 16% 18% Employer growth JS growth 0 1000 2000 3000 4000 5000 Applications per month 2015 2016 2017 3+Dimensions -Color
- 42. 3+Dimensions–Facets 45% 30% 21% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% campaign 1 campaign 2 campaign 3 80% 22% 44% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% campaign 1 campaign 2 campaign 3 33% 34% 21% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% campaign 1 campaign 2 campaign 3 Percentage of conversion per campaign Cairo Alexandria Aswan
- 43. 3+Dimensions–Facets 45% 30% 21% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% campaign 1 campaign 2 campaign 3 80% 22% 44% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% campaign 1 campaign 2 campaign 3 33% 34% 21% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% campaign 1 campaign 2 campaign 3 Percentage of conversion per campaign Cairo Alexandria Aswan Using colors to facilitate comparison across charts
- 44. 3+Dimensions–RadarChart 0 5 10 15 20 25 30 35 Distribution Chance creation RecoveryTake On Shooting Player A Player B Can Compare 2 items in multiple dimensions at the same time
- 45. Colors–Culturenorms • Red is warning/negative • Green is positive • The darker the color the more the value • Yellow/blue - Green/Red combinations are not color blind friendly 0 100 300 500 1000
- 46. Thetool–Difficultylevel Easy to use Google Data Studio programming skills required Data pre processing Capabilities High UX capabilities , easy integration with web
- 47. Thetool–Price$$ Free Google Data Studio Expensive Free for static chart but with hidden cost in hosting setup
- 48. Offline No connection to the data source Online Connected to the live data Difficulty Pricing UI Advanced charts Integration with web interface Interactivity Data pre processing Excel , Google sheets, Data wrapper Easy Free Good No No No No JS packages (D3 , High chart) Programming skills Free - there is enterprise edition Excellent Yes Easy Medium limited Programming packages (R, Python) Programming skills Free - there is enterprise edition V. Good Yes Medium Easy Yes BI tools Power BI, Tableau) Medium Expensive Excellent Yes No Easy No Glimpseintovisualizationtools
- 49. Truelies! There are three kinds of lies: Lies, Damn lies and Statistics
- 50. Truelies! Fox News showing the effect of Bush tax cuts Non Zero Base!!!
- 51. Truelies! 50% 55% 60% 65% 70% 75% 2016 2017 Our Platform Market Share in the past 2 years Starting from non zero base to give illusion of double growth 0% 20% 40% 60% 80% 100% 2016 2017 Our Platform Market Share in the past 2 years
- 52. Truelies! Inverted Axis to give the illusion of decline instead of increase
- 53. Truelies! Apple showing all time cumulative sales of iPhone
- 54. Truelies! Apple showing all time cumulative sales of iPhone The Actual values of sales
- 55. Truelies! 0 10000 20000 30000 40000 50000 60000 70000 80000 90000 100000 Jan Feb Mar Apr May June July Aug Sep Oct Cumulative Job posts per month Using cumulative value to fake increase 0 2000 4000 6000 8000 10000 12000 14000 Jan Feb Mar Apr May June July Aug Sep Oct Job posts per month
- 56. Truelies! Pie or Pizza?!! Values doesn’t sum to 100% No Comment!!
- 57. “WHY” is your first & most important question Bars are your best friend If categories are many… keep the pie for the dinner Keep it simple yet informative Be truthful to your audience In Summary
- 58. Questions?
- 59. Additional Resources • Brief History of Data visualization • How to design charts & graphs • Summary of The functional Art • Blog on turning bad charts to good ones • Table & Graph Design • Data Visualization Catalog • The Ethics of data visualization