The%20DMAIC%20Lean%20Six%20Sigma%20Project%20and%20Team%20Tools%20Approach%20%20Analyze%20Phase%20(Part%201)

1
The DMAIC Lean Six Sigma Project and Team Tools
Approach Analyze Phase(Part 1)
2
Lean Six Sigma Black Belt / Combo Training!
Analyze (Part 1) Agenda

• Welcome Back and Brief Review
• Analyze Overview
• Data and Basic Statistics
• Understanding Variation
• Descriptive Statistics
• Distributions and Analysis
• Normality
• Applications / Lessons Learned / Conclusions
• Next Steps

3
Lean Six Sigma DMAIC Phase Objectives

• Define what needs to be improved and why
• Measurewhat is the current state/performance
  level and potential causes
• Analyzecollect data and test to determine
  significant contributing causes
• Improveidentify and implement improvements for
  the significant causes
• Controlhold the gains of the improved process
  and monitor

4
What is Six Sigma?

• A high performance measure of excellence
• A metric for quality
• A business philosophy to improve customer
  satisfaction
• Focuses on processes and customers
• Delivers results that matter for all key
  stakeholders
• A tool for eliminating process variation
• Structured methodology to reduce defects
• Enables cultural change, it is transformational

5
Six Sigma Is a Set of Powerful Tools
Define Measure Analyze Improve Control
Problem Definition Process Mapping Key Factors (x) Selection Matrix OCAP
Project Management Cause Effect Matrix Basic Statistics Prioritization chart Standard Work
High Level Mapping Fishbone Diagram Regression FMEA Feedback Loops
Descriptive Statistics Statistical Analysis Hypothesis Testing Simulation Transition Plans
Pareto Value Stream Map ANOVA Future State Process Map Control Plans
Benchmarking MSA FMEA SPC
Cost/Benefit Analysis Capability Proportions Visual Control
6
Six Sigma applied effectively

• Increases customer satisfaction
• Lowers costs
• Builds better leaders
• Empowers an organization to be more data-driven

7
The Basic Philosophy of Lean Six Sigma

• All processes have variation and waste
• All variation and waste has causes
• Typically only a few causes are significant
• To the degree that those causes can be understood
  they can be controlled
• Designs must be robust to the effects of the
  remaining process variation
• This is true for products, processes, information
  transfer, transactions, everything
• Uncontrolled variation and waste is the enemy

8
The basic focus of Six Sigma

• The outputs (results) are a function of the
  inputs!
• Consistently meeting the needs of the customer is
  a function of how consistent /reliable the
  processes/inputs are that go into providing the
  service or product for the customer!
• KPOV f(KPIVs)
• Y f(Xs)

9
The basic focus of Six Sigma

• Therefore, to understand the output (results) we
  are getting, we must study and understand the
  process and inputs that go into producing the
  output we are getting.
• Y f(Xs)
• data-driven problem solving and continuous
  improvement

10
Yf(x)
x
x
x
x
x
x
x
Process MappingParetoCE Matrix,
FishboneFMEASIPOC Capability StudyMeasurement
Systems Analysis
X's
Measure Phase15-70 xs
Pareto Chart, Correlation/Regression Hypothesis
Tests, ANOVA, Descriptive Statistics, t-tests,
Proportions
Analyze Phase7-15 xs
Prioritization Matrix, Improvement Ideas, CE
Matrix, Future State Map, PDSA
Improve Phase 3-7 xs
Control Plans, SOPs, SPC, Mistake Proofing
Control Phase 3 or fewer xs
Only the Critical Xs need to be monitored and
controlled long term
11
Long-Term Yield vs Process Sigma
100
3.4 DPMO
233 DPMO
6,210 DPMO
Long-term Yield
66,807 DPMO
90
80
70
308,537 DPMO
60
50
40
30
690,000 DPMO
20
10
0
0
2
4
3
6
1
5
Process Sigma
12
Six Sigma AnalyzeIdentifying the Key Xs to
Improve the Process
13
Six Sigma Analyze Phase

• Make a habit of discussing a problem on the
  basis of the data and respecting the facts shown
  by them. - Kaoru Ishikawa

14
Six Sigma DMAIC Projects

• Analyze Phase
• What does the process data reveal?
• What are the Critical Key Xs?
• How much variation in Y from the Key Xs?
• What Xs can be and need to be improved (Root
  Causes)?

15
Analyze Objectives (pg. 12-14)

• Establish the capability of the process
• Establish an improvement goalthe performance
  objective
• Study the stability, shape, center, and spread of
  the process
• Determine the vital Xs that impact the project Y
• Make recommendations for the Improve phase
• Analyzecollect data and test to determine
  significant contributing causes

16
Six Sigma AnalyzeThe Data-Driven Approach
Process Analysis and Obvious Xs
17
Lean Six Sigma Project and Team Basic Tools

18
Lean Six Sigma Project and Team Basic Tools

• Process Flow Chart (pg. 33-44)
• A visual display of the key steps and flow of a
  process, also called a process map. Usually
  standard symbols are used to construct process
  flow charts. These include boxes (or rectangles)
  for specific steps, diamonds for decision points,
  ovals for defined starting and stopping points,
  and arrows to indicate flow.
• Processes can include providing a service, making
  or delivering products, information sharing,
  design, etc. Should represent the current as-is
  state of the process!

19
Process Flow Chart Lean Six Sigma Project
Selection
A Gap Exists
Define Potential Project
Draft Problem Statement
Identify the Metrics
Determine the Outputs (Y)
Two Or Fewer Outputs?
Redefine Project Scope
No
Reconsider Project
No
Yes
Meets Six Sigma Criteria?
Charter and Launch Project
Yes
Quantify the Opportunity
Calculate Benefits
20
Process AnalysisDetailed Analysis of Process
Delays or Errors

• Identifying process delays or potential errors is
  an important analyze phase activity. Going into
  greater detail in identifying the type and source
  of delay or error will help to more clearly
  define the root cause and thereby produce a more
  robust solution and overall improvement.

21
Analyze Roadmap Process AnalysisTypes of
Process Delays or Errors

• Gaps
• Redundancies
• Implicit or unclear requirements
• Bottlenecks
• Hand-offs
• Conflicting objectives
• Common problem areas

22
Lean Six Sigma Project and Team Basic Tools

23
Six Sigma AnalyzeData and Basic Statistics
The Use of Data to Make Decisions With an
Understanding of Variation
24
The Six Sigma Approach DMAIC Projects
Practical Problem
Statistical Problem
Statistical Solution
Practical Solution
Six Sigma applies statistical tools to practical
problems. The key is data-driven projects and
decision making. Improve toward near perfection
3.4 Defects per Million Opportunities (3.4 DPMO)
25
Understanding Variation

• I abhor averages. I like the individual case. A
  man may have six meals one day and none the next,
  making an average of three meals per day, but
  that is not a good way to live.
• Louis D. Brandeis

26
"Teen use Turns Upward"
high school seniors who smoke daily
1992 17.3 1993 19.0
Source USA Today, June 21, 1994
27
"Teen use Turns Upward"
high school seniors who smoke daily
1984 18.8 1985 19.6 1986 18.7 1987
18.6 1988 18.1 1989 18.9 1990 19.2 19
91 18.2 1992 17.3 1993 19.0
28
(No Transcript)
29
(No Transcript)
30
(No Transcript)
31
(No Transcript)
32
Bad News about Teen Smoking Steady Decline In
Teen Smoking Has Leveled Off, Study Finds
high school seniors who smoke daily
2008 11.4 2009 11.2
Source CBSnews.com, June 9, 2010
33
Bad News about Teen Smoking Steady Decline In
Teen Smoking Has Leveled Off, Study Finds
34
(No Transcript)
35
Understanding Variation
Percent Excellent - Taste, Temperature,
Variety April 2004
36
Understanding Variation
37
Understanding Variation

• All things vary. Probability allows us to
  determine if an event is common cause variation
  (random variation), or attributable to a specific
  cause or causes (special cause variation).

38
Understanding Variation (pg 118)

• Common cause variation variation due to random
  shifts in factors that are always present in the
  process
• Special cause variation variation above and
  beyond normal variation, arising from factors
  that are not always present in the process

39
Understanding Variation

• Managing special cause variation
• Find a data point that probably represents
  special cause variation (a statistical outlier)
• Track the root cause
• Eliminate the root cause
• Should result in a more stable, predictable
  process and smaller variation

Source IHC Institute Advanced Training Program
40

• Average TAT dropped by 10 Minutes
• Significant decrease in variability
• Fewer STAT Orders
• Not meeting 30 minute goal

25
Period Jan - May 2005
20
15
10
5
Mean54.42 Std Dev52.36 N102
0
0
3
6
9
2
2
2
3
3
3
3
4
4
1
1
1
0
0
0
1
4
7
0
3
6
9
2
5
2
5
8
0
0
0
0
0
0
0
0
0
0
0
0
Total Time (Minutes)
10
Period June - July 2005
8
6
4
Mean43.91 Std Dev17.75 N24
2
0
0
3
6
9
2
2
2
3
3
3
3
4
4
1
1
1
0
0
0
1
4
7
0
3
6
9
2
5
2
5
8
0
0
0
0
0
0
0
0
0
0
0
0
Total Time (Minutes)
41
Understanding Variation

• Managing common cause variation
• The level of random variation is a physical
  attribute of the process. Therefore, in order to
  reduce common cause variation, you must develop a
  new process with a new level of variation that is
  superior to the old process
• Often, the new process is a variation of the old
  process
• DMAIC to change the process

42
Understanding Variation
To achieve a new level of performance, examine
and improve the process...
Before
After
worse
better
Quality
worse
better
Quality
Source IHC Institute Advanced Training Program
43
Improvement in Cycle Time Right?Why do
we want to plot data over time?
Did we improve?
70
35
Provost, Lloyd. CHAI Fall Conference. Nashville,
TN. Sep 2004
44
Example 1
Cycle Time Results for Examples 1, 2 and 3
Example 2
Example 3
Provost, Lloyd. CHAI Fall Conference. Nashville,
TN. Sep 2004
45
Tampering

• Using special cause methods in an attempt to
  manage common cause (random) variation
• Tampering not only wastes time and effort, it
  also seriously harms process performance!

46
(No Transcript)
47
Troubled Waters Shark Kills Man, Leaves Woman
in Critical Condition Sept. 4 A shark ripped
off a man's leg, killing him, and mauled his
girlfriend, leaving her in critical condition, in
an attack off North Carolina's Outer Banks that
shocked doctors by its viciousness.
Emergency vehicles patrol the beach along North
Carolina's Outer Banks, where a man was killed
and his girlfriend mauled in a shark attack.
(ABCNEWS.com)
48
Florida Tops Shark Attack List Cynthia Mills,
Discovery.com News July 21, 2000 It's
summer, and the sharks are bitingwith a 50
percent increase in attacks six attacks already,
up from four total last July emphasis added.
49
Va. to Probe 2 Fatal Shark Attacks Wednesday,
September 05 BOB LEWIS Associated Press Writer
RICHMOND, Va. (AP) - Gov. Jim Gilmore created a
task force of experts Wednesday to investigate
the nation's two fatal shark attacks over the
Labor Day weekend.
50
Variation

• The power of statistical process control is
  DISCRIMINATION
• separates signal from noise
• a tool to help us know when to act
• a tool to help us know when NOT to act
  (tampering)
• Without understanding how much measurements vary
  naturally, it is impossible to understand the
  magnitude of the difference.

51
Six Sigma AnalyzeDescriptive Statistics
(Describing What the Data/Process Looks Like)
52
Random Sampling
X
X
X
X
X
X
X
X
X
X
X
X
Sample
X
X
X
Each element has an equal chance of being chosen
X
X
X
X
X
Population
53
Collecting output data over time results in a
distribution of the data There is spread of the
data, and a central tendency of the data
Distribution
Y
Sigma is the symbol for Standard Deviation
(measure of variation in the data)
54
Lean Six Sigma Project and Team Basic Tools

• Basic Statistics (pg. 104-110)
• When gathering and analyzing data, often we need
  to know something about location, or bunching of
  the data, and the amount of spread or variation.
  This leads us to two distinct types of measures
• Measures of Variation (Spread)
• Range, Standard Deviation
• Measures of Central Tendency
• Mean, Median, Mode

55
Data Distribution Analysis

• Terms you may hear regarding Spread
• Range is a simple measure of variation as it is
  the highest value (Xmax) minus the lowest value
  (Xmin) in the data set.
• Standard Deviation is a measure of variation and
  it is loosely defined as the average distance the
  individual values in the data set are varying
  from the mean of the data set. It is the square
  root of variance.

56
Standard Deviation (pg. 109-110)
Sample standard deviation
Population standard deviation
57
Six Sigma performance requires . . .
s
Distribution
s
Y
Z6
. . . plus/minus six sigmas within the output
spec
58
Use Six Sigma tools to reduce variation. . .
Cost of Quality Savings, Improved delivered
Quality
LSL
. . . reducing variability is the essence of Six
Sigma
59
Data Distribution Analysis

• Terms you may hear regarding Central Tendency
  
• Mean is another name for average. Sum all of the
  values in a data set and divide by the total
  number (n) of individual values in the set.
  
• Median is the absolute middle point (value) of a
  data set. For an even number (n), find the mean
  between the two middle values
• Mode is the value that occurs most frequently in
  a data set.

60
Data-Driven Problem Solving

• To be successful with any Six Sigma project, you
  have to affect positive/sustainable change in at
  least one of the following
• Reduce Variation
• Shift the Mean
• Eliminate Outliers

61
Characteristic of the Performance Gap
(Problem) Accuracy and/or Precision
Off-Target
Variation
On-Target
Center Process
Reduce Spread
LSL Lower spec limit USL Upper spec limit
The statistical approach to problem solving
62
Data Distribution Analysis

• Standard score or Z score
• The standard score simply indicates how many
  standard deviations a given value is above or
  below the distributions mean
• z value - mean / standard deviation
  

63
Data Distribution Analysis

• Standard score or Z score
• The calculated z score value is compared to a z
  table which gives the corresponding area under
  the curve (percentage of data)
  

64
Area under the curve
Where the z-score 0.00 50 of the distribution
is below the z-score
-4
-3
3
5
4
2
1
-5
-2
-1
0
65
Area under the curve
Where the z-score 1.00 84.13 of the
distribution is below the z-score
-4
-3
3
5
4
2
1
-5
-2
-1
0
66
Area under the curve
Where the z-score 2.00 97.72 of the
distribution is below the z-score
-4
-3
3
5
4
2
1
-5
-2
-1
0
67
Area under the curve
What percentage is above the z-score?
Where the z-score1.5
6.67
-4
-3
3
5
4
2
1
-5
-2
-1
0
93.33
1 .9333
68
Area between two limits
The area between -2s and 2s
95.44
-4
-3
3
5
4
2
1
-5
-2
-1
0
97.72
2.28
97.72 2.28
69
(No Transcript)
70
Performance over time. . .
. . . reducing variability is the essence of Six
Sigma
71
Data Distribution Analysis

• Terms you may hear regarding Distributions
• Skewness is a measure of the shape, and in
  particular the asymmetry, of a distribution
  
• Kurtosis is a measure of the peak shape of a
  probability distribution. A platykurtic
  distribution is flatter, while a leptokurtic
  distribution has a more acute peak.
  

72
Data Distribution Analysis

• Distributions Skewness
• Skewness - A negative value indicates skewness to
  the left, and a positive value indicates skewness
  to the right. The normal distribution has a
  skewness of zero. A zero skewness value though
  does not necessarily indicate perfect symmetry.
  

73
Skewed Distribution Curves
Positively skewed
-4
-3
3
5
4
2
1
-5
-2
-1
0
Negatively skewed
-4
-3
3
5
4
2
1
-5
-2
-1
0
74
Data Distribution Analysis

• Distributions Kurtosis
• Kurtosis is a measure of the peak shape of a
  probability distribution. A positive value
  indicates the distribution has a sharper peak
  (leptokurtic). A negative value indicates a
  flatter peak (platykurtic).

75
Kurtosis
Leptokurtic Positive excess kurtosis
-4
-3
3
-5
-1
5
4
2
1
-2
0
Platykurtic Negative excess kurtosis
-4
-3
3
5
4
2
1
-5
-2
-1
0
76
Is the data distribution normal? What does
being normal mean, and why is it important?
77
Does This Data Represent A Normal Distribution -
How do we know?
77
78
Data Distribution Analysis

• Normality Testing
• How well does a sample set of data fit a normal
  distribution?
• Visually
• Anderson-Darling test
• p-value

79
Data Distribution Analysis

• Normality Testing Visual
• Although it is not statistically valid to
  visually determine if a distribution is normal,
  it is usually a good practice and reference to
  visually display the data. Regardless of the
  shape of the distribution, we are often
  addressing at least one of the following issues
  
• Shifting the mean
• Reducing variation
• Eliminating / reducing significant outliers

80
Data Distribution Analysis

• Normality Testing Anderson-Darling
• A statistical test of whether there is evidence
  that a given sample of data comes from a given
  (normal) distribution. Typically, smaller
  Anderson-Darling (A2) values indicate a better
  fit of the data to a given distribution. The
  hypothesis of normality is generally rejected
  (alpha .05) at an A2 value of 0.75 or greater.
  

81
Data Distribution Analysis

• Normality Testing p-Value
• Think of the null hypothesis for normality
  testing as there is no difference in my sample
  data distribution and a normal distribution.
  Therefore, a normality test p-value of 0.05, or
  less (alpha .05) means that the null must go
  (reject). Your sample data can not be
  represented well by a normal distribution.
  

82
Penny exercise (Part 1)

• Take 20 pennies from the jar
• Randomly put them into 4 groups of 5
• Calculate the average year for each of your four
  samples (round to 1 decimal point)
• Randomly put them in groups of 10, and calculate
  the average year for the 2 samples (round to
  nearest 1 decimal point)

83
Penny exercise (Part 2)

• As a class, we will plot the averages in Minitab
  for N5 and N10
• What shape are the 2 distributions?
• What do the descriptive statistics tell us?

84
(No Transcript)
85
Central Limit Theorem (pg. 114)

• Many statistical tests or inferences apply only
  if the data are normally distributed.
• However, many continuous data sets are not
  normally distributed!
• So how do we deal with data that is not normally
  distributed?
• Option 1 Apply tests for non-normal
  distributions.
• Option 2 Apply the Central Limit Theorem.

86
Central Limit Theorem (p. 114)

• The Central Limit Theorem dictates that when you
  repeatedly calculate statistics for a process or
  characteristic, the repeated sample statistics
  have variation themselves.
• Simply, the distribution of all sample means is
  normal, as long as sample sizes are large enough.
  In addition, the mean distribution will have the
  same mean as the original population. Therefore,
  tests for normal distribution can be applied to
  the sample distribution.
• The more samples you take, and the larger the
  sample size, the variation will decrease.

87
Sampling Distributions
http//onlinestatbook.com/stat_sim/sampling_dist/i
ndex.html
88
Central Limit Theorem (pg. 114)

• If random samples (of size n) are repeatedly
  drawn from a population with a finite mean, when
  n is large
• the distribution of the means will be
  approximately normal
• the mean of that distribution is equal to the
  population mean
• the standard error decreases
• the standard deviation of the sample means
  decreases.
• NOTE If the parent population is normally
  distributed, the sample means will have an exact
  normal distribution, regardless of the sample
  size.

89
Six Sigma AnalyzeIdentifying the Key Xs to
fix that will most likely result in a
measureable improvement in the process Y (output)