"50 Shades of Truth: Visualisation and Communication of - - PowerPoint PPT Presentation

50 shades of truth visualisation and communication of
SMART_READER_LITE
LIVE PREVIEW

"50 Shades of Truth: Visualisation and Communication of - - PowerPoint PPT Presentation

www.imsd.ch "50 Shades of Truth: Visualisation and Communication of Uncertainty" Rodolphe Dewarrat & Vera Grttling, IMSD www.imsd.ch Issue Decision to use analytics Data Preprocessing Algorithms Raw


slide-1
SLIDE 1

"50 Shades of Truth: Visualisation and Communication of Uncertainty" Rodolphe Dewarrat & Vera Gärttling, IMSD

www.imsd.ch

slide-2
SLIDE 2
  • Issue
  • Decision to use analytics
  • Data
  • Preprocessing
  • Algorithms
  • Raw results
  • Interpretation of results
  • Action

www.imsd.ch

slide-3
SLIDE 3

Available Data and Truth:

www.imsd.ch

percentage of relevant information available

slide-4
SLIDE 4

Even the Best Algorithms Available

  • don’t make the difference between these situations
  • don’t know what information could be interesting
  • are based on violated assumptions

www.imsd.ch

Visit the booth of the swiss statistical society

slide-5
SLIDE 5

The best analytical method ADF

www.imsd.ch

Presented on the 28th of May at an expert round table of statisticians at the ETH Developed at the University of Exeter (UK), tested and validated by international experts

slide-6
SLIDE 6

interface

  • Trained for millenaries to recognize patterns

The best analytical method ADF

www.imsd.ch

slide-7
SLIDE 7
  • Today probably our best brain to machine

interface

  • Trained for millenaries to recognize patterns

instantly

The human visual system

www.imsd.ch

slide-8
SLIDE 8

Example 1 of 4

Aircraft section plot www.imsd.ch

slide-9
SLIDE 9

Example 2 of 4

www.imsd.ch

slide-10
SLIDE 10

0.25 0.5 0.75 1

Example 3 of 4

Uncertain radar plot

2 4 6 8

satisfaction

Employees Customers Suppliers Stock holders Management

www.imsd.ch

satisfaction

Employees Customers Suppliers Stock holders Management

slide-11
SLIDE 11

Example 4 of 4

SC plot

FR

VD

BE ZH 1 1cm2 = 10k customers churn canton

100% 80% 60% 40%

FR VD JU BE ZH

www.imsd.ch

JU

100% 80% 60% 40%

FR VD BE JU ZU

10k 80k 60k 40k

FR VD BE JU ZU

slide-12
SLIDE 12

Conclusion

  • Before starting anything (e.g. buying a data cluster, doing an analysis)

look at your data and consider its validity and truth.

  • This step cannot be replaced by a machine.
  • Strongly consider working with partners and tools (software, team)

that allow you a high flexibility in your data representation. Thank you for your attention contact: rodolphe.dewarrat@imsd.ch twitter: @imsd_analytics www.imsd.ch