The gap between theory and practice of SPC Eugenio Epprecht PUCRio, Brazil

1
The gap between theory and practice of
SPCEugenio Epprecht - PUC-Rio, Brazil
2

• IN PRACTICE
• many (most?) processes are
• autocorrelated
• and/or
• multivariate or multi-channel
• or just not the simplest case

3

• IN THEORY
• there are several techniques for SPC for process
  of such nature (in research papers and
• also in textbooks),
• but...

4

• IN PRACTICE
• The person or team in charge of SPC
• only knows classic Shewharts charts (for the
  univariate, i.i.d. case)
• even for univariate and uncorrelated processes
  often applies the charts in a wrong way
• and...

5

• even when the chart is correctly applied, design
  is arbitrary no consideration of efficiency
  issues (not even knowledge abouy false-alarm
  probabilities, charts power, etc.)
• (IN THEORY,
• there is a lot of works on how to make your SPC
  more efficient)

6

• In summary
• Wrong technique
• Right technique, applied wrongly
• Right technique, applied correctly but
  inefficiently

7

• What is the origin of these problems?
• local problem?
• not enough undergraduate courses in
  Statistics and SPC?
• more is needed than undergraduate courses?
• (particular twist of each practical problem)
• attitude? company culture?
• ??????

8

• How to fix them?
• more hours of undergraduate Statistics and SPC
  courses?
• more in-house training?
• companies should hire statisticians?
• intensifying the interaction between
  universities and companies between researchers
  and practitioners?
• ??????
• NOTE companies want simple solutions for complex
  (or not too simple) problems