Transcript Process Improvement (Even a CEO can learn it)

Process Improvement with Solitaire
(So easy, even a
can do it)
Using the PC Solitaire game to learn
basic (and advanced) techniques of
Process Improvement
Presented by: Mark Berron, CQE, CSQE, 6ơ BB
Contact: [email protected]
Benefits of using Solitaire
• It gives the user a hands-on experience of
creating and using data.
• The processes are simple
• The processes can be improved
• The process improvement can be
measured.
• Even Six Sigma techniques can be learned
• It’s easy!!!
What we’ll do today…
•
•
•
•
•
•
•
•
•
What are the Solitaire Requirements??
Process mapping
Play Solitaire
Brainstorming on Improvements
Play more Solitaire
Measure the improvements
Translating this to your work processes
What this can mean to your bottom line…
For further information…
Requirements
Definitions:
New Card
Upper Deck (UD)
Playing Deck (PD)
Lower Deck (LD)
Columns
Rules (Defined)
•
•
•
•
•
Objective is to get highest score (No time limits). Technically this is putting all cards into the Upper Deck by their
respective suits.
Basic process is: New cards go to Upper Deck. If can’t be played there, they are played on Lower Deck in alternative
colors in descending order. If they can’t be played on the Lower Deck, then a new Card is chosen.
The lowest Card in a Column in the lower Deck must be moved to the Upper Deck if possible and before choosing a
New Card. When a column in the Lower Deck becomes empty, only a King can be moved there.
Every card from the Lower Deck must be played until all options exhausted. Then a New Card is chosen.
When there are no more cards in the Playing Deck, or if all cards are in the Upper Deck, the game is over.
Process Map
• What is a process map? ISO?
• Here’s the process map for Solitaire:
Choose
New
Card
Any
cards
left?
N
End Game
Record
score
Y
Move it
to Upper
Deck?
Y
Put in UD
N
Move it
to Lower
Deck?
Y
N
Play Rest of LD to UD
Open game
Put in LD
Now, let’s play
•
•
•
•
•
•
Use standard type of play
Las Vegas Scoring
No timing
One card at a time
Don’t think, just do.
And observe. Maybe take notes.
Value Stream Mapping
Based on what you have seen, what does each step contribute
to the process? Let’s make a Value Stream Map (VSM) from
the previous Process Map…
Choose
New
Card
Any
cards
left?
N
End Game
Record
score
Y
10
Move it
to Upper
Deck?
Y
Put in UD
N
5
Move it
to Lower
Deck?
N
Y
Put in LD
10
Play Rest of LD to UD
Open game
Brainstorming
• Based on what you have seen, what are
possible improvements??
Here are some ideas I have:
• Do another pass?
• Instead of moving new cards directly to
Upper Deck, first move them to Lower
Deck.
• Let cards move from column to column
• Let cards go from upper to lower deck
• Don’t play out all cards before New Card
Which ideas do we try?
• What is potential of each improvement?
• How is it measured?
• How much does it cost?
Improvement potential
First, what’s the baseline?:
The one pass.
I chose four improvements:
1. Second pass
2. Let new cards go down to columns before
going to upper deck
3. Allow cards to be moved to any column on
bottom deck
4. Allow cards to be returned to columns from top
finished deck.
VSM 1st improvement
Add the second pass to the process…
Choose
New
Card
Any
cards
left?
Y
N
N
Y
Put in UD
N
End Game
Click on PD?
Record
score
-100
10
Move it
to Upper
Deck?
5
Move it
to Lower
Deck?
Y
N
Y
Put in LD
10
Play Rest of LD to UD
Open game
Measuring the improvements
Basically, two ways to measure:
1. Play out each improvement over a
number of rounds.
2. Play out all improvements over a known
design (DOE).
The 2nd choice saves time and money, but is
it accurate? Will it tell you which
improvement is really best?
The long way…
• Here are the results of going through each
improvement 100 times, with each adding on to
the next. Thus number 2, has both 1 and 2
improvements included.
• For every improvement, scores were recorded
for both one pass and two passes.
Some data…..
Some typical data gathering…
Some basic charts
• Trend lines
– Mean
– Standard deviation
– Number of wins
– Average winning score
Tracking the mean… Trend Chart
Means over 4 experiments
270
80
73
70
250
65
60
60
Score
50
210
40
37
30
190
Mean difference
230
Mean 1
Mean 2
Mean diff
20
170
10
150
0
1
2
3
Experiment
4
Mean 1 = One pass
Mean 2 = Two
Standard Deviation and Means
2nd pass higher Scores and Means
Number of high scores actually goes down, but average is higher. Why?
2nd pass winning scores vs. mean
What does it mean???
The summary data is:
Means over 4 experiments
270
80
73
70
250
65
60
60
Score
50
210
40
37
30
190
Mean difference
230
Mean 1
Mean 2
Mean diff
20
170
10
150
0
1
2
3
Experiment
4
Mean 1 = One pass
Mean 2 = Two
The 1st improvement appears to be the most significant
The 2nd improvement appears to be the 2nd most significant
The 4th improvement appears to be the significant on the first pass
The second pass increases improvements except in 4th case
Mean 1 Factors show some effects of interaction.
What about DOE??
Results across the variables
Pareto Chart of DOE
250
200
150
100
50
Series1
-100
-150
AC
D
BD
D
C
AB
D
AC
D
AB
C
AD
AB
BC
D
C
D
B
AB
C
-50
BC
A
0
Comparing DOE to actuals
• Both show A as the best option.
• Actuals show B as 2nd best while 3rd in
DOE.
• Actuals show C as a small or negative
influence, while it ranks only 6th in DOE.
• Actuals show D with a medium influence,
while DOE shows it as fourth.
• DOE shows a big BC interaction, but the
actuals show none or negative.
More basic analysis
What else can be learned?
• Cost/Benefit: Compute cost of the
improvement by assessing the added
time. How does the cost compare to
added score over time?
• Risk: Assess the risk of taking on an
improvement. The 2nd pass creates a lot
more volatility. Is it worth it??
• What is the probability of success on the
2nd pass? How would you figure that out?
Compare to the real world…
• What process do you have at work that
would benefit from process improvement.
• What process needs a VSM?
• Why not use ISO process as the baseline
for process improvement? Or CMMI?
• Brainstorm for improvements.
• Find out which has the most weight… the
low hanging fruit. How much will it cost?
• Then do it!!!!
What other techniques can be
learned from Solitaire?
•
•
•
•
•
ANOVA
X-Bar and R charts
Probability
Ishikawa (Cause and effect diagrams)
Taguchi, Latin Square, partial factorial
designs
• 5S (Clean up the process)
• And more….
Xbar-R Chart
X-bar R Chart
Further information
• I am in the process of creating a training
manual along with an Excel-based data
collection/analysis program. It should be
complete by end of August 2007. If you
might be interested in a follow up session
or the materials themselves, please let me
know and I will forward a version to you.
My email address is: [email protected]