2555571 six sigma and minitab 13
TRANSCRIPT
-
8/3/2019 2555571 Six Sigma and Minitab 13
1/396
QSM 754SIX SIGMA APPLICATIONS
AGENDA
-
8/3/2019 2555571 Six Sigma and Minitab 13
2/396
Research and Benchmarking Web Sites
CommercialQuality Progresshttp://www.qualityprogress.asq.org
Minitabhttp://www.minitab.com
General Electric Annual Report 1997http://www.ge.com/annual97/sixsigma/
W. Edwards Deming Institute
http://www.deming.org/deminghtml/wedi.html
AcademicUniversity of Edinburgh
http://www.ucs.ed.ac.uk/usd/socsci/stats/minitab.html
http://www.qualityprogress.asq.org/http://www.minitab.com/http://www.ge.com/annual97/sixsigma/http://www.deming.org/deminghtml/wedi.htmlhttp://www.ucs.ed.ac.uk/usd/socsci/stats/minitab.htmlhttp://www.ucs.ed.ac.uk/usd/socsci/stats/minitab.htmlhttp://www.deming.org/deminghtml/wedi.htmlhttp://www.ge.com/annual97/sixsigma/http://www.minitab.com/http://www.qualityprogress.asq.org/ -
8/3/2019 2555571 Six Sigma and Minitab 13
3/396
Day 1 Agenda
Welcome and Introductions Course Structure
Meeting Guidelines/Course Agenda/Report Out Criteria
Group Expectations
Introduction to Six Sigma Applications
Red Bead Experiment
Introduction to Probability Distributions
Common Probability Distributions and Their Uses
Correlation Analysis
-
8/3/2019 2555571 Six Sigma and Minitab 13
4/396
Day 2 Agenda
Team Report Outs on Day 1 Material
Central Limit Theorem
Process Capability
Multi-Vari Analysis
Sample Size Considerations
-
8/3/2019 2555571 Six Sigma and Minitab 13
5/396
Day 3 Agenda
Team Report Outs on Day 2 Material
Confidence Intervals
Control Charts
Hypothesis Testing
ANOVA (Analysis of Variation)
Contingency Tables
-
8/3/2019 2555571 Six Sigma and Minitab 13
6/396
Day 4 Agenda
Team Report Outs on Practicum Application
Design of Experiments
Wrap Up - Positives and Deltas
-
8/3/2019 2555571 Six Sigma and Minitab 13
7/396
Class Guidelines
Q&A as we go Breaks Hourly
Homework Readings As assigned in Syllabus
Grading Class Preparation 30%
Team Classroom Exercises 30%
Team Presentations 40%
10 Minute Daily Presentation (Day 2 and 3) on Application of previous days work
20 minute final Practicum application (Last day)
Copy on Floppy as well as hard copy
Powerpoint preferred Rotate Presenters
Q&A from the class
-
8/3/2019 2555571 Six Sigma and Minitab 13
8/396
INTRODUCTION TO SIXSIGMA APPLICATIONS
-
8/3/2019 2555571 Six Sigma and Minitab 13
9/396
Learning Objectives
Have a broad understanding of statistical concepts andtools.
Understand how statistical concepts can be used toimprove business processes.
Understand the relationship between the curriculum and
the four step six sigma problem solving process(Measure, Analyze, Improve and Control).
-
8/3/2019 2555571 Six Sigma and Minitab 13
10/396
What is Six Sigma?
A Philosophy
A Quality Level
A Structured Problem-Solving Approach
A Program
Customer Critical To Quality (CTQ) Criteria Breakthrough Improvements Fact-driven, Measurement-based, Statistically Analyzed
Prioritization Controlling the Input & Process Variations Yields a Predictable
Product
6s = 3.4 Defects per Million Opportunities
Phased Project: Measure, Analyze, Improve, Control
Dedicated, Trained BlackBelts Prioritized Projects
Teams - Process Participants & Owners
-
8/3/2019 2555571 Six Sigma and Minitab 13
11/396
POSITIONING SIX SIGMATHE FRUIT OF SIX SIGMA
Ground FruitLogic and Intuition
Low Hanging FruitSeven Basic Tools
Bulk of FruitProcess Characterizationand Optimization
ProcessEntitlement
Sweet FruitDesign for Manufacturability
-
8/3/2019 2555571 Six Sigma and Minitab 13
12/396
UNLOCKING THE HIDDEN FACTORY
VALUESTREAM TO
THECUSTOMER
PROCESSES WHICHPROVIDE PRODUCT VALUEIN THE CUSTOMERS EYES
FEATURES ORCHARACTERISTICS THECUSTOMER WOULD PAYFOR.
WASTE DUE TOINCAPABLEPROCESSES
WASTE SCATTERED THROUGHOUTTHE VALUE STREAM
EXCESS INVENTORY
REWORK WAIT TIME EXCESS HANDLING EXCESS TRAVEL DISTANCES TEST AND INSPECTION
Waste is a significant cost driver and has a majorimpact on the bottom line...
-
8/3/2019 2555571 Six Sigma and Minitab 13
13/396
-
8/3/2019 2555571 Six Sigma and Minitab 13
14/396
The Focus of Six Sigma..
Y = f(x)
All critical characteristics (Y)are driven by factors (x) whichare upstream from the
results.
Attempting to manage results(Y) only causes increasedcosts due to rework, test andinspection
Understanding and controllingthe causative factors (x) is thereal key to high quality at lowcost...
-
8/3/2019 2555571 Six Sigma and Minitab 13
15/396
INSPECTION EXERCISE
The necessity of training farm hands for first classfarms in the fatherly handling of farm livestock isforemost in the minds of farm owners. Since the
forefathers of the farm owners trained the farm handsfor first class farms in the fatherly handling of farmlivestock, the farm owners feel they should carry onwith the family tradition of training farm hands of first
class farms in the fatherly handling of farm livestockbecause they believe it is the basis of goodfundamental farm management.
How many fs can you identify in 1 minute of inspection.
-
8/3/2019 2555571 Six Sigma and Minitab 13
16/396
INSPECTION EXERCISE
The necessity of* training f*arm hands f*or f*irst classf*arms in the f*atherly handling of* f*arm livestock isf*oremost in the minds of* f*arm owners. Since the
f*oref*athers of* the f*arm owners trained the f*armhands f*or f*irst class f*arms in the f*atherly handlingof* f*arm livestock, the f*arm owners f*eel they shouldcarry on with the f*amily tradition of* training f*arm
hands of* f*irst class f*arms in the f*atherly handlingof* f*arm livestock because they believe it is the basisof* good f*undamental f*arm management.
How many fs can you identify in 1 minute of inspection.36 total are available.
-
8/3/2019 2555571 Six Sigma and Minitab 13
17/396
-
8/3/2019 2555571 Six Sigma and Minitab 13
18/396
IMPROVEMENT ROADMAP
Breakthrough
Strategy
Characterization
Phase1:
Measurement
Phase 2:
Analysis
Optimization
Phase 3:
Improvement
Phase4:
Control
Define the problem andverify the primary andsecondary measurementsystems.
Identify the few factorswhich are directlyinfluencing the problem.
Determine values for thefew contributing factors
which resolve theproblem.
Determine long termcontrol measures whichwill ensure that thecontributing factors
remain controlled.
Objective
-
8/3/2019 2555571 Six Sigma and Minitab 13
19/396
Measurements are critical...
If we cant accurately measure
something, we really dont know much
about it.
If we dont know much about it, wecant control it.
If we cant control it, we are at the
mercy of chance.
-
8/3/2019 2555571 Six Sigma and Minitab 13
20/396
WHY STATISTICS?THE ROLE OF STATISTICS IN SIX SIGMA..
WE DONT KNOW WHAT WE DONT KNOW IF WE DONT HAVE DATA, WE DONT KNOW
IF WE DONT KNOW, WE CAN NOT ACT
IF WE CAN NOT ACT, THE RISK IS HIGH
IF WE DO KNOW AND ACT, THE RISK IS MANAGED
IF WE DO KNOW AND DO NOT ACT, WE DESERVE THE LOSS.
DR. Mikel J. Harry
TO GET DATA WE MUST MEASURE
DATA MUST BE CONVERTED TO INFORMATION
INFORMATION IS DERIVED FROM DATA THROUGHSTATISTICS
USLT
LSL
-
8/3/2019 2555571 Six Sigma and Minitab 13
21/396
WHY STATISTICS?THE ROLE OF STATISTICS IN SIX SIGMA..
Ignorance is not bliss, it is the food of failure andthe breeding ground for loss.
DR. Mikel J. Harry
Years ago a statistician might have claimed that
statistics dealt with the processing of data. Todays statistician will be more likely to say thatstatistics is concerned with decision making in theface of uncertainty.
Bartlett
USLT
LSL
-
8/3/2019 2555571 Six Sigma and Minitab 13
22/396
Sales Receipts
On Time Delivery
Process Capacity
Order Fulfillment Time
Reduction of WasteProduct Development Time
Process Yields
Scrap Reduction
Inventory Reduction
Floor Space Utilization
WHAT DOES IT MEAN?
Random Chance or Certainty.
Which would you choose.?
-
8/3/2019 2555571 Six Sigma and Minitab 13
23/396
Learning Objectives
Have a broad understanding of statistical concepts andtools.
Understand how statistical concepts can be used toimprove business processes.
Understand the relationship between the curriculum and
the four step six sigma problem solving process(Measure, Analyze, Improve and Control).
-
8/3/2019 2555571 Six Sigma and Minitab 13
24/396
RED BEAD EXPERIMENT
-
8/3/2019 2555571 Six Sigma and Minitab 13
25/396
Learning Objectives
Have an understanding of the difference betweenrandom variation and a statistically significant event.
Understand the difference between attempting tomanage an outcome (Y) as opposed to managingupstream effects (xs).
Understand how the concept of statistical significancecan be used to improve business processes.
-
8/3/2019 2555571 Six Sigma and Minitab 13
26/396
WELCOME TO THE WHITE BEADFACTORY
HIRING NEEDS
BEADS ARE OUR BUSINESS
PRODUCTION SUPERVISOR
4 PRODUCTION WORKERS
2 INSPECTORS
1 INSPECTION SUPERVISOR
1 TALLY KEEPER
-
8/3/2019 2555571 Six Sigma and Minitab 13
27/396
STANDING ORDERS
Follow the process exactly.
Do not improvise or vary from the documented process.
Your performance will be based solely on your ability to producewhite beads.
No questions will be allowed after the initial training period.
Your defect quota is no more than 5 off color beads allowed perpaddle.
-
8/3/2019 2555571 Six Sigma and Minitab 13
28/396
WHITE BEAD MANUFACTURING PROCESSPROCEDURES The operator will take the bead paddle in the right hand.
Insert the bead paddle at a 45 degree angle into the bead bowl. Agitate the bead paddle gently in the bead bowl until all spaces are
filled.
Gently withdraw the bead paddle from the bowl at a 45 degree angleand allow the free beads to run off.
Without touching the beads, show the paddle to inspector #1 and waituntil the off color beads are tallied.
Move to inspector #2 and wait until the off color beads are tallied.
Inspector #1 and #2 show their tallies to the inspection supervisor. Ifthey agree, the inspection supervisor announces the count and thetally keeper will record the result. If they do not agree, the inspectionsupervisor will direct the inspectors to recount the paddle.
When the count is complete, the operator will return all the beads tothe bowl and hand the paddle to the next operator.
-
8/3/2019 2555571 Six Sigma and Minitab 13
29/396
INCENTIVE PROGRAM
Low bead counts will be rewarded with a bonus.
High bead counts will be punished with a reprimand.
Your performance will be based solely on your abilityto produce white beads.
Your defect quota is no more than 7 off color beadsallowed per paddle.
-
8/3/2019 2555571 Six Sigma and Minitab 13
30/396
PLANT RESTRUCTURE
Defect counts remain too high for the plant to beprofitable.
The two best performing production workers will beretained and the two worst performing productionworkers will be laid off.
Your performance will be based solely on your abilityto produce white beads.
Your defect quota is no more than 10 off color beadsallowed per paddle.
-
8/3/2019 2555571 Six Sigma and Minitab 13
31/396
OBSERVATIONS.
WHAT OBSERVATIONS DID YOU
MAKE ABOUT THIS PROCESS.?
-
8/3/2019 2555571 Six Sigma and Minitab 13
32/396
The Focus of Six Sigma..
Y = f(x)
All critical characteristics (Y)are driven by factors (x) whichare downstream from the
results.
Attempting to manage results(Y) only causes increasedcosts due to rework, test andinspection
Understanding and controllingthe causative factors (x) is thereal key to high quality at lowcost...
-
8/3/2019 2555571 Six Sigma and Minitab 13
33/396
Learning Objectives
Have an understanding of the difference betweenrandom variation and a statistically significant event.
Understand the difference between attempting tomanage an outcome (Y) as opposed to managingupstream effects (xs).
Understand how the concept of statistical significancecan be used to improve business processes.
-
8/3/2019 2555571 Six Sigma and Minitab 13
34/396
INTRODUCTION TO
PROBABILITYDISTRIBUTIONS
-
8/3/2019 2555571 Six Sigma and Minitab 13
35/396
Learning Objectives
Have a broad understanding of what probabilitydistributions are and why they are important.
Understand the role that probability distributions play indetermining whether an event is a random occurrence orsignificantly different.
Understand the common measures used to characterize apopulation central tendency and dispersion.
Understand the concept of Shift & Drift.
-
8/3/2019 2555571 Six Sigma and Minitab 13
36/396
Why do we Care?
An understanding ofProbability Distributions isnecessary to:
Understand the concept anduse of statistical tools.
Understand the significance ofrandom variation in everydaymeasures.
Understand the impact ofsignificance on the successfulresolution of a project.
-
8/3/2019 2555571 Six Sigma and Minitab 13
37/396
IMPROVEMENT ROADMAPUses of Probability Distributions
Breakthrough
Strategy
Characterization
Phase1:
Measurement
Phase 2:
Analysis
Optimization
Phase 3:
Improvement
Phase4:
Control
Establish baseline datacharacteristics.
Project Uses
Identify and isolatesources of variation.
Use the concept of shift &drift to establish projectexpectations.
Demonstrate before andafter results are not
random chance.
-
8/3/2019 2555571 Six Sigma and Minitab 13
38/396
Focus on understanding the concepts
Visualize the concept
Dont get lost in the math.
KEYS TO SUCCESS
-
8/3/2019 2555571 Six Sigma and Minitab 13
39/396
Measurements are critical...
If we cant accurately measure
something, we really dont know much
about it.
If we dont know much about it, wecant control it.
If we cant control it, we are at the
mercy of chance.
-
8/3/2019 2555571 Six Sigma and Minitab 13
40/396
Types of Measures
Measures where the metric is composed of aclassification in one of two (or more) categories is calledAttribute data. This data is usually presented as acount or percent.
Good/Bad Yes/No
Hit/Miss etc.
Measures where the metric consists of a number whichindicates a precise value is called Variable data. Time
Miles/Hr
-
8/3/2019 2555571 Six Sigma and Minitab 13
41/396
COIN TOSS EXAMPLE
Take a coin from your pocket and toss it 200 times.
Keep track of the number of times the coin falls asheads.
When complete, the instructor will ask you for yourhead count.
-
8/3/2019 2555571 Six Sigma and Minitab 13
42/396
COIN TOSS EXAMPLE
130120110100908070
10000
5000
0
Cum
ulativeFrequency
Results from 10,000 people doing a coin toss 200 times.Cumulative Count
130120110100908070
600
500
400
300
200
100
0
"Head Count"
Frequency
Results from 10,000 people doing a coin toss 200 times.Count Frequency
130120110100908070
100
50
0
"Head Count"
Cumu
lativePercent
Results from 10,000 people doing a coin toss 200 times.Cumulative Percent
Cumulative Frequency
CumulativePercent
Cumulative count is simply the total frequencycount accumulated as you move from left to
right until we account for the total population of10,000 people.
Since we know how many people were in thispopulation (ie 10,000), we can divide each of thecumulative counts by 10,000 to give us a curvewith the cumulative percent of population.
-
8/3/2019 2555571 Six Sigma and Minitab 13
43/396
COIN TOSS PROBABILITY EXAMPLE
130120110100908070
100
50
0
Cumu
lativePercent
Results from 1 0,000 peop le doing a coin toss 200 timesCumulative Percent
This means that we can nowpredict the change thatcertain values can occur
based on these percentages.Note here that 50% of thevalues are less than ourexpected value of 100.
This means that in a futureexperiment set up the sameway, we would expect 50%of the values to be less than100.
-
8/3/2019 2555571 Six Sigma and Minitab 13
44/396
COIN TOSS EXAMPLE
130120110100908070
600
500
400
300
200
100
0
"Head Count"
Frequency
Results from 10,000 people doing a coin toss 200 times.Count Frequency
130120110100908070
100
50
0
"Head Count"
Cumu
lativePercent
Results from 10,000 people doing a coin toss 200 times.Cumulative Percent
We can now equate a probability to theoccurrence of specific values or groups ofvalues.
For example, we can see that the
occurrence of a Head count of less than74 or greater than 124 out of 200 tossesis so rare that a single occurrence wasnot registered out of 10,000 tries.
On the other hand, we can see that thechance of getting a count near (or at) 100
is much higher. With the data that wenow have, we can actually predict each ofthese values.
-
8/3/2019 2555571 Six Sigma and Minitab 13
45/396
COIN TOSS PROBABILITY DISTRIBUTION
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
NUMBER OF HEADS
PROCESS CENTEREDON EXPECTED VALUE
s
SIGMA (s ) IS A MEASUREOF SCATTER FROM THE
EXPECTED VALUE THAT
CAN BE USED TOCALCULATE A PROBABILITY
OF OCCURRENCE
SIGMA VALUE (Z)
CUM % OF POPULATION
58 65 72 79 86 93 100 107 114 121 128 135 142
.003 .135 2.275 15.87 50.0 84.1 97.7 99.86 99.997
130120110100908070
600
500
400
300
200
100
0
Frequency
If we know wherewe are in the
population we canequate that to aprobability value.This is the purposeof the sigma value(normal data).
% of population = probability of occurrence
-
8/3/2019 2555571 Six Sigma and Minitab 13
46/396
-
8/3/2019 2555571 Six Sigma and Minitab 13
47/396
Probability and Statistics
the odds of Colorado University winning the nationaltitle are 3 to 1
Drew Bledsoes pass completion percentage for the last6 games is .58% versus .78% for the first 5 games
The Senator will win the election with 54% of the popularvote with a margin of +/- 3%
Probability and Statistics influence our lives daily Statistics is the universal lanuage for science Statistics is the art of collecting, classifying,presenting, interpreting and analyzing numericaldata, as well as making conclusions about thesystem from which the data was obtained.
-
8/3/2019 2555571 Six Sigma and Minitab 13
48/396
Population Vs. Sample (Certainty Vs. Uncertainty)
A sample is just a subset of all possible values
population
sample
Since the sample does not contain all the possible values,
there is some uncertainty about the population. Hence any
statistics, such as mean and standard deviation, are just
estimates of the true population parameters.
-
8/3/2019 2555571 Six Sigma and Minitab 13
49/396
-
8/3/2019 2555571 Six Sigma and Minitab 13
50/396
Inferential Statistics
Inferential Statistics is the branch of statistics that deals withdrawing conclusions about a population based on informationobtained from a sample drawn from that population.
While descriptive statistics has been taught for centuries,inferential statistics is a relatively new phenomenon havingits roots in the 20th century.
We infer something about a population when only informationfrom a sample is known.
Probability is the link betweenDescriptive and Inferential Statistics
-
8/3/2019 2555571 Six Sigma and Minitab 13
51/396
WHAT DOES IT MEAN?
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
NUMBER OF HEADS
s
SIGMA VALUE (Z)
CUM % OF POPULATION
58 65 72 79 86 93 100 107 114 121 128 135 142
.003 .135 2.275 15.87 50.0 84.1 97.7 99.86 99.997
130120110100908070
600
500
400
300
200
100
0
Fre
quency
And the first 50trials showedHead Counts
greater than 130?
WHAT IF WE MADE A CHANGE TO THE PROCESS?
Chances are verygood that theprocess distributionhas changed. Infact, there is aprobability greater
than 99.999% thatit has changed.
-
8/3/2019 2555571 Six Sigma and Minitab 13
52/396
USES OF PROBABILITY DISTRIBUTIONS
CriticalValue
CriticalValue
CommonOccurrence
RareOccurrence
RareOccurrence
Primarily these distributions are used to test for significant differences in data sets.
To be classified as significant, the actual measured value must exceed a criticalvalue. The critical value is tabular value determined by the probability distributionand the risk of error. This risk of error is called a risk and indicates the probabilityof this value occurring naturally. So, an a risk of .05 (5%) means that this critical
value will be exceeded by a random occurrence less than 5% of the time.
-
8/3/2019 2555571 Six Sigma and Minitab 13
53/396
SO WHAT MAKES A DISTRIBUTION UNIQUE?
CENTRAL TENDENCY
Where a population is located.
DISPERSION
How wide a population is spread.
DISTRIBUTION FUNCTION
The mathematical formula thatbest describes the data (we willcover this in detail in the nextmodule).
-
8/3/2019 2555571 Six Sigma and Minitab 13
54/396
COIN TOSS CENTRAL TENDENCY
1 3 01 2 01 1 01 0 09 08 07 0
6 0 0
5 0 0
4 0 0
3 0 0
2 0 0
1 0 0
0
Num
berofoccurrences
What are some of the ways that we can easily indicatethe centering characteristic of the population?
Three measures have historically been used; the
mean, the median and the mode.
-
8/3/2019 2555571 Six Sigma and Minitab 13
55/396
WHAT IS THE MEAN?
ORDERED DATA SET
-5
-3
-1
-1
0
0
0
0
0
1
3
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 4
The mean has already been used in several earlier modulesand is the most common measure of central tendency for apopulation. The mean is simply the average value of thedata.
n=12
xi = - 2
mean xx
n
i= = =-
= - 212
17.
Mean
-
8/3/2019 2555571 Six Sigma and Minitab 13
56/396
WHAT IS THE MEDIAN?
ORDERED DATA SET
-5
-3
-1
-1
0
0
0
0
0
1
3
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 4
If we rank order (descending or ascending) the data set forthis distribution we could represent central tendency by theorder of the data points.
If we find the value half way (50%) through the data points, wehave another way of representing central tendency. This iscalled the median value.
Median
Value
Median
50% of datapoints
-
8/3/2019 2555571 Six Sigma and Minitab 13
57/396
WHAT IS THE MODE?
ORDERED DATA SET
-5
-3
-1
-1
0
0
0
0
0
1
3
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 4
If we rank order (descending or ascending) the data set forthis distribution we find several ways we can represent centraltendency.
We find that a single value occurs more often than any other.Since we know that there is a higher chance of this
occurrence in the middle of the distribution, we can use thisfeature as an indicator of central tendency. This is called themode.
Mode Mode
-
8/3/2019 2555571 Six Sigma and Minitab 13
58/396
MEASURES OF CENTRAL TENDENCY, SUMMARY
MEAN ( )(Otherwise known as the average)
XX
ni= =
-= 2
1217.
X
ORDERED DATA SET
-5
-3
-1
-1
0
0
00
0
1
3
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 4
ORDERED DATA SET
-5
-3
-1
-1
0
0
0
0
0
1
3
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 4
ORDERED DATA SET
-5
-3
-1
-1
0
0
00
0
1
3-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 4
MEDIAN
(50 percentile data point)
Here the median value falls between two zerovalues and therefore is zero. If the values were
say 2 and 3 instead, the median would be 2.5.MODE
(Most common value in the data set)
The mode in this case is 0 with 5 occurrenceswithin this data.
Mediann=12
n/2=6
n/2=6
} Mode = 0Mode = 0
-
8/3/2019 2555571 Six Sigma and Minitab 13
59/396
SO WHATS THE REAL DIFFERENCE?
MEAN
The mean is the mostconsistently accurate measure ofcentral tendency, but is more
difficult to calculate than theother measures.
MEDIAN AND MODE
The median and mode are both
very easy to determine. Thatsthe good news.The bad news
is that both are more susceptibleto bias than the mean.
-
8/3/2019 2555571 Six Sigma and Minitab 13
60/396
SO WHATS THE BOTTOM LINE?
MEAN
Use on all occasions unless acircumstance prohibits its use.
MEDIAN AND MODE
Only use if you cannot usemean.
-
8/3/2019 2555571 Six Sigma and Minitab 13
61/396
COIN TOSS POPULATION DISPERSION
1 3 01 2 01 1 01 0 09 08 07 0
6 0 0
5 0 0
4 0 0
3 0 0
2 0 0
1 0 0
0
Num
berofoccurrence
s
What are some of the ways that we can easily indicate the dispersion(spread) characteristic of the population?
Three measures have historically been used; the range, the standard
deviation and the variance.
-
8/3/2019 2555571 Six Sigma and Minitab 13
62/396
WHAT IS THE RANGE?
ORDERED DATA SET-5
-3
-1
-1
00
0
0
0
13
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 4
The range is a very common metric which is easilydetermined from any ordered sample. To calculate the rangesimply subtract the minimum value in the sample from themaximum value.
Range
RangeMaxMin
Range x xMAX MIN= - = - - =4 5 9( )
-
8/3/2019 2555571 Six Sigma and Minitab 13
63/396
WHAT IS THE VARIANCE/STANDARD DEVIATION?
The variance (s2) is a very robust metric which requires a fair amount of work todetermine. The standard deviation(s) is the square root of the variance and is themost commonly used measure of dispersion for larger sample sizes.
( )s
X X
n
i2
2
1
61 67
12 15 6=
-
-=
-=
..
DATA SET-5
-3-1-10000013
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 4
XX
n
i= =
-=
212
-.17X Xi -
-5-(-.17)=-4.83
-3-(-.17)=-2.83
-1-(-.17)=-.83
-1-(-.17)=-.83
0-(-.17)=.17
0-(-.17)=.17
0-(-.17)=.17
0-(-.17)=.17
0-(-.17)=.17
1-(-.17)=1.17
3-(-.17)=3.17
4-(-.17)=4.17
( )X Xi
-
2
(-4.83)2=23.32
(-2.83)2=8.01
(-.83)2=.69
(-.83)2=.69
(.17)2=.03
(.17)2=.03
(.17)2=.03
(.17)2=.03
(.17)2=.03
(1.17)2=1.37
(3.17)2=10.05
(4.17)2=17.39
61.67
-
8/3/2019 2555571 Six Sigma and Minitab 13
64/396
MEASURES OF DISPERSION
RANGE (R)(The maximum data value minus the minimum)
ORDERED DATA SET
-5-3
-1
-1
0
0
0
0
0
1
3
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 4
ORDERED DATA SET
-5
-3
-1
-1
0
0
00
0
1
3
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 4
VARIANCE (s2)
(Squared deviations around the center point)
STANDARD DEVIATION (s)
(Absolute deviation around the center point)
Min=-5
R X X= - = - - =max min ( )4 6 10
Max=4DATA SET
-5
-3
-1
-10
0
0
00
1
3
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 4
XX
n
i= =
-= 2
12-.17
( )s
X X
ni2
2
1
61 67
12 15 6=
--
=-
= . .
X Xi --5-(-.17)=-4.83
-3-(-.17)=-2.83
-1-(-.17)=-.83
-1-(-.17)=-.83
0-(-.17)=.17
0-(-.17)=.17
0-(-.17)=.17
0-(-.17)=.17
0-(-.17)=.17
1-(-.17)=1.17
3-(-.17)=3.174-(-.17)=4.17
( )X Xi-
2
(-4.83)2=23.32
(-2.83)2=8.01
(-.83)2=.69
(-.83)2=.69
(.17)2=.03
(.17)2=.03
(.17)2=.03
(.17)2=.03
(.17)2=.03
(1.17)2=1.37
(3.17)2=10.05(4.17)2=17.39
61.67
s s= = =2 5 6 2 37. .
-
8/3/2019 2555571 Six Sigma and Minitab 13
65/396
SAMPLE MEAN AND VARIANCE EXAMPLE
$= =
XN
Xi
s ( )$22
1= =
-2s
n
-X Xi
Xi10151214
109
111210
12
1
2
3
4
5
6
7
8
910
SX
Xi
-X Xi ( )2-X Xi
2s
-
8/3/2019 2555571 Six Sigma and Minitab 13
66/396
SO WHATS THE REAL DIFFERENCE?
VARIANCE/ STANDARD DEVIATION
The standard deviation is the mostconsistently accurate measure ofcentral tendency for a single
population. The variance has theadded benefit of being additive overmultiple populations. Both are difficultand time consuming to calculate.
RANGEThe range is very easy to determine.Thats the good news.The bad news
is that it is very susceptible to bias.
-
8/3/2019 2555571 Six Sigma and Minitab 13
67/396
SO WHATS THE BOTTOM LINE?
VARIANCE/ STANDARDDEVIATION
Best used when you haveenough samples (>10).
RANGE
Good for small samples (10 orless).
-
8/3/2019 2555571 Six Sigma and Minitab 13
68/396
SO WHAT IS THIS SHIFT & DRIFT STUFF...
The project is progressing well and you wrap it up. 6 monthslater you are surprised to find that the population has taken a
shift.
-12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12
USLLSL
-
8/3/2019 2555571 Six Sigma and Minitab 13
69/396
SO WHAT HAPPENED?
All of our work was focused in a narrow time frame.Over time, other long term influences come and gowhich move the population and change some of itscharacteristics. This is called shift and drift.
Historically, this shift and driftprimarily impacts the position ofthe mean and shifts it 1.5 s fromits original position.
Original Study
-
8/3/2019 2555571 Six Sigma and Minitab 13
70/396
VARIATION FAMILIES
Variation is presentupon repeatmeasurements withinthe same sample.
Variation is presentupon measurements ofdifferent samples
collected within a shorttime frame.
Variation is presentupon measurementscollected with a
significant amount oftime between samples.
Sources ofVariation
Within IndividualSample
Piece to
Piece
Time to Time
-
8/3/2019 2555571 Six Sigma and Minitab 13
71/396
SO WHAT DOES IT MEAN?
To compensate for these longterm variations, we mustconsider two sets of metrics.
Short term metrics are thosewhich typically are associatedwith our work. Long term metricstake the short term metric dataand degrade it by an average of1.5s.
-
8/3/2019 2555571 Six Sigma and Minitab 13
72/396
IMPACT OF 1.5s SHIFT AND DRIFTZ PPM ST Cpk PPM LT (+1.5 s)
0.0 500,000 0.0 933,193
0.1 460,172 0.0 919,243
0.2 420,740 0.1 903,199
0.3 382,089 0.1 884,930
0.4 344,578 0.1 864,334
0.5 308,538 0.2 841,3450.6 274,253 0.2 815,940
0.7 241,964 0.2 788,145
0.8 211,855 0.3 758,036
0.9 184,060 0.3 725,747
1.0 158,655 0.3 691,462
1.1 135,666 0.4 655,4221.2 115,070 0.4 617,911
1.3 96,801 0.4 579,260
1.4 80,757 0.5 539,828
1.5 66,807 0.5 500,000
1.6 54,799 0.5 460,172
1.7 44,565 0.6 420,740
Here, you can see that theimpact of this concept is
potentially very significant. Inthe short term, we have driventhe defect rate down to 54,800
ppm and can expect to seeoccasional long term ppm tobe as bad as 460,000 ppm.
-
8/3/2019 2555571 Six Sigma and Minitab 13
73/396
SHIFT AND DRIFT EXERCISE
We have just completed a project and have presented thefollowing short term metrics:
Zst=3.5
PPMst=233
Cpkst=1.2
Calculate the long
term values for eachof these metrics.
-
8/3/2019 2555571 Six Sigma and Minitab 13
74/396
Learning Objectives
Have a broad understanding of what probabilitydistributions are and why they are important.
Understand the role that probability distributions play in
determining whether an event is a random occurrence orsignificantly different.
Understand the common measures used to characterize a
population central tendency and dispersion.
Understand the concept of Shift & Drift.
-
8/3/2019 2555571 Six Sigma and Minitab 13
75/396
COMMON PROBABILITY
DISTRIBUTIONS ANDTHEIR USES
-
8/3/2019 2555571 Six Sigma and Minitab 13
76/396
Learning Objectives
Have a broad understanding of how probabilitydistributions are used in improvement projects.
Review the origin and use of common probabilitydistributions.
-
8/3/2019 2555571 Six Sigma and Minitab 13
77/396
-
8/3/2019 2555571 Six Sigma and Minitab 13
78/396
IMPROVEMENT ROADMAPUses of Probability Distributions
Breakthrough
Strategy
Characterization
Phase1:
Measurement
Phase 2:
Analysis
Optimization
Phase 3:
Improvement
Phase4:
Control
Baselining Processes
Verifying Improvements
Common Uses
-
8/3/2019 2555571 Six Sigma and Minitab 13
79/396
Focus on understanding the use of the distributions
Practice with examples wherever possible
Focus on the use and context of the tool
KEYS TO SUCCESS
PROBABILITY DISTRIBUTIONS WHERE
-
8/3/2019 2555571 Six Sigma and Minitab 13
80/396
X
X X X
X
Data points vary, but as the data accumulates, it forms a distribution which occurs naturally.
Location Spread Shape
Distributions can vary in:
PROBABILITY DISTRIBUTIONS, WHEREDO THEY COME FROM?
COMMON PROBABILITY DISTRIBUTIONS
-
8/3/2019 2555571 Six Sigma and Minitab 13
81/396
-4
-3
-2
-1
0
1
2
3
4
0 1 2 3 4 5 6 7
-4
-3
-2
-1
0
1
2
3
4
0 1 2 3 4 5 6 7
0
1
2
3
4
0 1 2 3 4 5 6 7
Original Population
Subgroup Average
Subgroup Variance (s2)
Continuous Distribution
Normal Distribution
c2
Distribution
COMMON PROBABILITY DISTRIBUTIONS
-
8/3/2019 2555571 Six Sigma and Minitab 13
82/396
THE LANGUAGE OF MATH
Symbol Name Statistic Meaning Common Uses
a Alpha Significance level Hypothesis Testing,DOE
c2 Chi Square Probability Distribution Confidence Intervals, ContingencyTables, Hypothesis Testing
S Sum Sum of Individual values Variance Calculationst t, Student t Probability Distribution Hypothesis Testing, Confidence Interval
of the Mean
n Sample
Size
Total size of the Sample
Taken
Nearly all Functions
Nu Degree of Freedom Probability Distributions, HypothesisTesting, DOE
Beta Beta Risk Sample Size Determination Delta Difference between
population means
Sample Size Determination
SigmaValue
Number of Standard
Deviations a value Exists
from the Mean
Probability Distributions, Process
Capability, Sample Size Determinations
Population and Sample Symbology
-
8/3/2019 2555571 Six Sigma and Minitab 13
83/396
Population and Sample Symbology
Value Population Sample
Mean Variance s2 s2Standard Deviation s sProcess Capability Cp
Binomial Mean
x
P P
Cp
-
8/3/2019 2555571 Six Sigma and Minitab 13
84/396
THREE PROBABILITY DISTRIBUTIONS
tX
s
n
CALC = -
Significant t tCALC CRIT=
Significant F FCALC CRIT= F sscalc = 12
22
( )c
a,df
e a
e
f f
f
2
2
=-
SignificantCALC CRIT
= c c2 2
Note that in each case, a limit has been established to determine what israndom chance verses significant difference. This point is called the criticalvalue. If the calculated value exceeds this critical value, there is very lowprobability (P
-
8/3/2019 2555571 Six Sigma and Minitab 13
85/396
-1s +1s
68.26%
+/- 1s = 68%2 tail = 32%
1 tail = 16%
-2s +2s
95.46%
+/- 2s = 95%2 tail = 4.6%
1 tail = 2.3%
-3s +3s
99.73%
+/- 3s = 99.7%2 tail = 0.3%1 tail = .15%
Common Test Values
Z(1.6) = 5.5% (1 tail a=.05)Z(2.0) = 2.5% (2 tail a=.05)
Z TRANSFORM
-
8/3/2019 2555571 Six Sigma and Minitab 13
86/396
The Focus of Six Sigma..
Y = f(x)
All critical characteristics (Y)are driven by factors (x) whichare downstream from the
results.
Attempting to manage results(Y) only causes increasedcosts due to rework, test andinspection
Understanding and controllingthe causative factors (x) is thereal key to high quality at lowcost...
Probability distributions identify sourcesof causative factors (x). These can beidentified and verified by testing whichshows their significant effects againstthe backdrop of random noise.
BUT WHAT DISTRIBUTION
-
8/3/2019 2555571 Six Sigma and Minitab 13
87/396
SHOULD I USE?
CharacterizePopulation
PopulationAverage
PopulationVariance
DetermineConfidence
Interval forPoint Values
F Stat (n>30)
F Stat (n5)
Z Stat (n>30)
Z Stat (p)
t Stat (n30)
Z Stat (p)
t Stat (,n
-
8/3/2019 2555571 Six Sigma and Minitab 13
88/396
These interactions form a newpopulation which can now be
used to predict futureperformance.
HOW DO POPULATIONS INTERACT?
-
8/3/2019 2555571 Six Sigma and Minitab 13
89/396
HOW DO POPULATIONS INTERACT?ADDING TWO POPULATIONS
Population means interact in a simple intuitive manner.
1 2
Means Add
1 + 2 = new
Population dispersions interact in an additive manner
s1 s2
Variations Add
s12 + s22 = snew2
snew
new
HOW DO POPULATIONS INTERACT?
-
8/3/2019 2555571 Six Sigma and Minitab 13
90/396
HOW DO POPULATIONS INTERACT?SUBTRACTING TWO POPULATIONS
Population means interact in a simple intuitive manner.
1 2
Means Subtract
1 - 2 = new
Population dispersions interact in an additive manner
s1 s2
Variations Add
s12 + s22 = snew2
snew
new
-
8/3/2019 2555571 Six Sigma and Minitab 13
91/396
TRANSACTIONAL EXAMPLE
Orders are coming in with the following characteristics:
Shipments are going out with the followingcharacteristics:
Assuming nothing changes, what percent of the time willshipments exceed orders?
X = $53,000/week
s = $8,000
X = $60,000/weeks = $5,000
TRANSACTIONAL EXAMPLE
-
8/3/2019 2555571 Six Sigma and Minitab 13
92/396
To solve this problem, we must create a new distribution to model the situation posedin the problem. Since we are looking for shipments to exceed orders, the resulting
distribution is created as follows:
X X Xshipments orders shipments orders- = - = - =$60 , $53, $7 ,000 000 000
( ) ( )s s sshipments orders shipments orders- = + = + =2 22 2
5000 8000 $9434
The new distribution looks like this with a mean of $7000 and astandard deviation of $9434. This distribution represents theoccurrences of shipments exceeding orders. To answer the originalquestion (shipments>orders) we look for $0 on this new distribution.Any occurrence to the right of this point will represent shipments >orders. So, we need to calculate the percent of the curve that existsto the right of $0.
$7000
$0
Shipments > orders
X = $53,000 in orders/week
s = $8,000
X = $60,000 shipped/week
s = $5,000
Orders Shipments
TRANSACTIONAL EXAMPLE
-
8/3/2019 2555571 Six Sigma and Minitab 13
93/396
TRANSACTIONAL EXAMPLE, CONTINUED
X X Xshipments orders shipments orders- = - = - =$60 , $53, $7 ,000 000 000
( ) ( )s s sshipments orders shipments orders- = + = + =2 22 2
5000 8000 $9434
To calculate the percent of the curve to the right of $0 we need toconvert the difference between the $0 point and $7000 into sigma
intervals. Since we know every $9434 interval from the mean is onesigma, we can calculate this position as follows:
$7000
$0
Shipments > orders
074
-=
-=
X
ss
$0 $7000
$9434.
Look up .74s in the normal table and you will find .77. Therefore, the answer to the original
question is that 77% of the time, shipments will exceed orders.
Now, as a classroom exercise, what percent of the time willshipments exceed orders by $10,000?
-
8/3/2019 2555571 Six Sigma and Minitab 13
94/396
MANUFACTURING EXAMPLE
2 Blocks are being assembled end to end and significantvariation has been found in the overall assembly length.
The blocks have the following dimensions:
Determine the overall assembly length and standarddeviation.
X1 = 4.00 inches
s1 = .03 inches
X2 = 3.00 inches
s2 = .04 inches
-
8/3/2019 2555571 Six Sigma and Minitab 13
95/396
-
8/3/2019 2555571 Six Sigma and Minitab 13
96/396
CORRELATION ANALYSIS
-
8/3/2019 2555571 Six Sigma and Minitab 13
97/396
Learning Objectives
Understand how correlation can be used to demonstratea relationship between two factors.
Know how to perform a correlation analysis and calculate
the coefficient of linear correlation (r).
Understand how a correlation analysis can be used in animprovement project.
-
8/3/2019 2555571 Six Sigma and Minitab 13
98/396
Why do we Care?
Correlation Analysis isnecessary to:
show a relationship betweentwo variables. This also sets the
stage for potential cause andeffect.
IMPROVEMENT ROADMAP
-
8/3/2019 2555571 Six Sigma and Minitab 13
99/396
IMPROVEMENT ROADMAPUses of Correlation Analysis
Breakthrough
Strategy
Characterization
Phase1:
Measurement
Phase 2:
Analysis
Optimization
Phase 3:
Improvement
Phase4:
Control
Determine and quantifythe relationship betweenfactors (x) and outputcharacteristics (Y)..
Common Uses
KEYS TO SUCCESS
-
8/3/2019 2555571 Six Sigma and Minitab 13
100/396
Always plot the data
Remember: Correlation does not always imply cause & effect
Use correlation as a follow up to the Fishbone Diagram
Keep it simple and do not let the tool take on a life of its own
WHAT IS CORRELATION?
-
8/3/2019 2555571 Six Sigma and Minitab 13
101/396
WHAT IS CORRELATION?
Input or x variable (independent)
Output or yvariable
(dependent)
Correlation
Y= f(x)
As the input variable changes,
there is an influence or biason the output variable.
WHAT IS CORRELATION?
-
8/3/2019 2555571 Six Sigma and Minitab 13
102/396
A measurable relationship between two variable data characteristics.
Not necessarily Cause & Effect (Y=f(x))
Correlation requires paired data sets (ie (Y1,x1), (Y2,x2), etc)
The input variable is called the independent variable (x or KPIV) since it isindependent of any other constraints
The output variable is called the dependent variable (Y or KPOV) since itis (theoretically) dependent on the value of x.
The coefficient of linear correlation r is the measure of the strength ofthe relationship.
The square of r is the percent of the response (Y) which is related to theinput (x).
WHAT IS CORRELATION?
-
8/3/2019 2555571 Six Sigma and Minitab 13
103/396
CALCULATING r
-
8/3/2019 2555571 Six Sigma and Minitab 13
104/396
Coefficient of Linear Correlation
Calculate sample covariance( )
Calculate sx and sy for eachdata set
Use the calculated values tocompute rCALC.
Add a + for positive correlationand - for a negative correlation.
( )( )s
x x y y
nxyi i
=- -
-
1
rs sCALCsxy
x y
=
sxy
While this is the most precise method to calculatePearsons r, there is an easier way to come up with a fairlyclose approximation...
APPROXIMATING r
-
8/3/2019 2555571 Six Sigma and Minitab 13
105/396
Coefficient of Linear Correlation
Y=f(x)
x
rW
L -
1
r - - = -16 7
12 647.
..
+ = positive slope
- = negative slope
L| | | | | | | | | | | | | | |1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
6.7 12.6
Plot the data on orthogonal axisDraw an Oval around the data
Measure the length and widthof the Oval
Calculate the coefficient of
linear correlation (r) based onthe formulas below
-
8/3/2019 2555571 Six Sigma and Minitab 13
106/396
HOW DO I KNOW WHEN I HAVE CORRELATION?
OrderedPairs rCRIT
5 .88
6 .81
7 .75
8 .71
9 .67
10 .63
15 .51
20 .44
25 .40
30 .3650 .28
80 .22
100 .20
The answer should strike a familiar cord at this pointWe have confidence (95%) that we have correlationwhen |rCALC|> rCRIT.
Since sample size is a key determinate of rCRIT we needto use a table to determine the correct rCRIT given thenumber of ordered pairs which comprise the complete
data set.
So, in the preceding example we had 60 ordered pairsof data and we computed a rCALC of -.47. Using the tableat the left we determine that the rCRIT value for 60 is .26.
Comparing |rCALC
|> rCRIT
we get .47 > .26. Therefore thecalculated value exceeds the minimum critical valuerequired for significance.
Conclusion: We are 95% confident that the observedcorrelation is significant.
L i Obj i
-
8/3/2019 2555571 Six Sigma and Minitab 13
107/396
Learning Objectives
Understand how correlation can be used to demonstratea relationship between two factors.
Know how to perform a correlation analysis and calculate
the coefficient of linear correlation (r).
Understand how a correlation analysis can be used in ablackbelt story.
-
8/3/2019 2555571 Six Sigma and Minitab 13
108/396
CENTRAL LIMIT THEOREM
L i Obj ti
-
8/3/2019 2555571 Six Sigma and Minitab 13
109/396
Learning Objectives
Understand the concept of the Central Limit Theorem.
Understand the application of the Central Limit Theoremto increase the accuracy of measurements.
-
8/3/2019 2555571 Six Sigma and Minitab 13
110/396
IMPROVEMENT ROADMAP
-
8/3/2019 2555571 Six Sigma and Minitab 13
111/396
IMPROVEMENT ROADMAPUses of the Central Limit Theorem
Breakthrough
Strategy
Characterization
Phase1:
Measurement
Phase 2:
Analysis
Optimization
Phase 3:
Improvement
Phase4:
Control
The Central LimitTheorem underlies all
statistic techniques whichrely on normality as afundamental assumption
Common Uses
KEYS TO SUCCESS
-
8/3/2019 2555571 Six Sigma and Minitab 13
112/396
Focus on the practical application of the concept
-
8/3/2019 2555571 Six Sigma and Minitab 13
113/396
WHAT IS THE CENTRAL LIMIT THEOREM?
Central Limit Theorem
For almost all populations, the sampling distribution of themean can be approximated closely by a normal distribution,
provided the sample size is sufficiently large.
Normal
Why do we Care?
-
8/3/2019 2555571 Six Sigma and Minitab 13
114/396
What this means is that no matter what kindof distribution we sample, if the sample sizeis big enough, the distribution for the meanis approximately normal.
This is the key link that allows us to use
much of the inferential statistics we havebeen working with so far.
This is the reason that only a few probabilitydistributions (Z, t and c2) have such broadapplication.
If a random event happens a great many times,the average results are likely to be predictable.
Jacob Bernoulli
HOW DOES THIS WORK?
-
8/3/2019 2555571 Six Sigma and Minitab 13
115/396
ParentPopulation
n=2
n=5
n=10
As you average a larger
and larger number ofsamples, you can see howthe original sampledpopulation is transformed..
ANOTHER PRACTICAL ASPECT
-
8/3/2019 2555571 Six Sigma and Minitab 13
116/396
snxx=
sThis formula is for the standard error of the mean.
What that means in layman's terms is that thisformula is the prime driver of the error term of themean. Reducing this error term has a direct impacton improving the precision of our estimate of themean.
The practical aspect of all this is that if you want to improve the precision of anytest, increase the sample size.
So, if you want to reduce measurement error (for example) to determine a betterestimate of a true value, increase the sample size. The resulting error will bereduced by a factor of . The same goes for any significance testing.Increasing the sample size will reduce the error in a similar manner.
1n
DICE EXERCISE
-
8/3/2019 2555571 Six Sigma and Minitab 13
117/396
Break into 3 teamsTeam one will be using 2 diceTeam two will be using 4 diceTeam three will be using 6 dice
Each team will conduct 100 throws of their dice andrecord the average of each throw.
Plot a histogram of the resulting data.
Each team presents the results in a 10 min report out.
Learning Objectives
-
8/3/2019 2555571 Six Sigma and Minitab 13
118/396
Learning Objectives
Understand the concept of the Central Limit Theorem.
Understand the application of the Central Limit Theoremto increase the accuracy of measurements.
-
8/3/2019 2555571 Six Sigma and Minitab 13
119/396
PROCESS CAPABILITY
ANALYSIS
Learning Objectives
-
8/3/2019 2555571 Six Sigma and Minitab 13
120/396
Learning Objectives
Understand the role that process capability analysisplays in the successful completion of an improvementproject.
Know how to perform a process capability analysis.
Why do we Care?
-
8/3/2019 2555571 Six Sigma and Minitab 13
121/396
Why do we Care?
Process Capability Analysis isnecessary to:
determine the area of focuswhich will ensure successfulresolution of the project.
benchmark a process to enabledemonstrated levels ofimprovement after successfulresolution of the project.
demonstrate improvement aftersuccessful resolution of theproject.
IMPROVEMENT ROADMAP
-
8/3/2019 2555571 Six Sigma and Minitab 13
122/396
IMPROVEMENT ROADMAPUses of Process Capability Analysis
Breakthrough
Strategy
Characterization
Phase1:
Measurement
Phase 2:
Analysis
Optimization
Phase 3:
Improvement
Phase4:
Control
Baselining a processprimary metric (Y) prior tostarting a project.
Common Uses
Characterizing thecapability of causitivefactors (x).
Characterizing a processprimary metric after
changes have beenimplemented todemonstrate the level ofimprovement.
KEYS TO SUCCESS
-
8/3/2019 2555571 Six Sigma and Minitab 13
123/396
Must have specification limits - Use process targets if no specs available
Dont get lost in the math
Relate to Z for comparisons (Cpk x 3 = Z)
For Attribute data use PPM conversion to Cpk and Z
WHAT IS PROCESS CAPABILITY?
-
8/3/2019 2555571 Six Sigma and Minitab 13
124/396
Process capability is simply a measure of how good a metric is performingagainst and established standard(s). Assuming we have a stable processgenerating the metric, it also allows us to predict the probability of the metricvalue being outside of the established standard(s).
Spec
Out of SpecIn Spec
Probability
Spec(Lower)
Spec(Upper)
In Spec Out of SpecOut of Spec
ProbabilityProbability
Upper and Lower Standards(Specifications)
Single Standard(Specification)
WHAT IS PROCESS CAPABILITY?
-
8/3/2019 2555571 Six Sigma and Minitab 13
125/396
Process capability (Cpk) is a function of how the population is centered (|-spec|) and the population spread (s).
Process Center
(|-spec|)
Spec(Lower)
Spec(Upper)
In SpecOut ofSpec
Out ofSpec
High Cpk Poor CpkSpec
(Lower)
Spec(Upper)
In Spec
Out ofSpec
Out ofSpec
Process Spread
(s)
Spec(Lower)
Spec(Upper)
In Spec
Out ofSpec
Out ofSpec
Spec(Lower)
Spec(Upper)
In Spec
Out ofSpec
Out ofSpec
-
8/3/2019 2555571 Six Sigma and Minitab 13
126/396
HOW IS PROCESS CAPABILITY CALCULATED
Spec(LSL)
Spec(USL)
Note:
LSL = Lower Spec Limit
USL = Upper Spec Limit
Distance between thepopulation mean andthe nearest spec limit(|-USL |). Thisdistance divided by3s is Cpk.
( )C
MIN LSL USL
PK=
- -
s
,
3
Expressed mathematically, this looks like:
PROCESS CAPABILITY EXAMPLE
-
8/3/2019 2555571 Six Sigma and Minitab 13
127/396
PROCESS CAPABILITY EXAMPLE
Calculation Values:
Upper Spec value = $200,000 maximumNo Lower Spec
= historical average = $250,000s = $20,000
Calculation:
Answer: Cpk= -.83
We want to calculate the process capability for our inventory. Thehistorical average monthly inventory is $250,000 with a standarddeviation of $20,000. Our inventory target is $200,000 maximum.
( ) ( )C
MIN LSL USL
PK =
- -
=
-
=
s
, $200, $250,
* $20, -.833
000 000
3 000
-
8/3/2019 2555571 Six Sigma and Minitab 13
128/396
Learning Objectives
-
8/3/2019 2555571 Six Sigma and Minitab 13
129/396
Learning Objectives
Understand the role that process capability analysisplays in the successful completion of an improvementproject.
Know how to perform a process capability analysis.
-
8/3/2019 2555571 Six Sigma and Minitab 13
130/396
MULTI-VARI ANALYSIS
Learning Objectives
-
8/3/2019 2555571 Six Sigma and Minitab 13
131/396
Learning Objectives
Understand how to use multi-vari charts in completing animprovment project.
Know how to properly gather data to construct multi-vari
charts.
Know how to construct a multi-vari chart.
Know how to interpret a multi-vari chart.
Why do we Care?
-
8/3/2019 2555571 Six Sigma and Minitab 13
132/396
Why do we Care?
Multi-Vari charts are a:
Simple, yet powerful way tosignificantly reduce the numberof potential factors which couldbe impacting your primary metric.
Quick and efficient method tosignificantly reduce the time andresources required to determinethe primary components of
variation.
IMPROVEMENT ROADMAPUses of Multi-Vari Charts
-
8/3/2019 2555571 Six Sigma and Minitab 13
133/396
Breakthrough
Strategy
Characterization
Phase1:
Measurement
Phase 2:Analysis
Optimization
Phase 3:
Improvement
Phase4:
Control
Eliminate a large numberof factors from theuniverse of potentialfactors.
Common Uses
KEYS TO SUCCESS
-
8/3/2019 2555571 Six Sigma and Minitab 13
134/396
Gather data by systematically sampling the existing process
Perform ad hoc training on the tool for the team prior to use
Ensure your sampling plan is complete prior to gathering data
Have team members (or yourself) do the sampling to avoid bias
Careful planning before you start
A World of Possible Causes (KPIVs)..
-
8/3/2019 2555571 Six Sigma and Minitab 13
135/396
The Goal of the logical search
is to narrow down to 5-6 key variables !
technique
tooling
operator error
humidity
temperature
supplier hardness
lubrication
old machinery
wrong spec.tool wear
handling damagepressure heat treat
fatigue
-
8/3/2019 2555571 Six Sigma and Minitab 13
136/396
REDUCING THE POSSIBILITIES
-
8/3/2019 2555571 Six Sigma and Minitab 13
137/396
How many guesses do you think it will take to find asingle word in the text book?
Lets try and see.
REDUCING THE POSSIBILITIES
-
8/3/2019 2555571 Six Sigma and Minitab 13
138/396
How many guesses do you think it will take to find a singleword in the text book?
Statistically it should take no more than 17 guesses217=2X2X2X2X2X2X2X2X2X2X2X2X2X2X2X2X2= 131,072
Most Unabridged dictionaries have 127,000 words.
Reduction of possibilities can be an extremely powerfultechnique..
PLANNING A MULTI-VARI ANALYSIS
-
8/3/2019 2555571 Six Sigma and Minitab 13
139/396
Determine the possible families of variation.Determine how you will take the samples.
Take a stratified sample (in order of creation).
DO NOT take random samples.
Take a minimum of 3 samples per group and 3 groups.The samples must represent the full range of the process.
Does one sample or do just a few samples stand out?There could be a main effect or an interaction at the cause.
MULTI VARI ANALYSIS VARIATION FAMILIES
-
8/3/2019 2555571 Six Sigma and Minitab 13
140/396
MULTI-VARI ANALYSIS, VARIATION FAMILIES
Variation is presentupon repeatmeasurements within
the same sample.
Variation is presentupon measurements ofdifferent samples
collected within a shorttime frame.
Variation is presentupon measurementscollected with a
significant amount oftime between samples.
Sources ofVariation
Within IndividualSample
Piece toPiece
Time to Time
The key is reducing the number of possibilities to a manageable few.
MULTI-VARI ANALYSIS, VARIATION SOURCES
-
8/3/2019 2555571 Six Sigma and Minitab 13
141/396
Within Individual Sample
Measurement Accuracy
Out of Round
Irregularities in Part
Piece to Piece
Machine fixturing
Mold cavity differences
Time to Time
Material Changes
Setup Differences
Tool Wear
Calibration Drift
Operator Influence
Manufacturing
(Machining)
Transactional
(Order
Rate)
Piece to Piece
Customer Differences
Order Editor
Sales Office
Sales Rep
Within Individual Sample
Measurement Accuracy
Line Item Complexity
Time to Time
Seasonal Variation
Management Changes
Economic Shifts
Interest Rate
HOW TO DRAW THE CHART
-
8/3/2019 2555571 Six Sigma and Minitab 13
142/396
Range within asingle sample
Sample 1
Average within asingle sample
Plot the first sample range with a point for themaximum reading obtained, and a point for the
minimum reading. Connect the points and plota third point at the average of the within samplereadings
Step 1
Range betweentwo sampleaverages
Sample 1
Sample 2Sample 3
Plot the sample ranges for the remaining piece
to piece data. Connect the averages of the
within sample readings.
Step 2
Plot the time to time groups in the same
manner.
Time 1Time 2
Time 3
Step 3
READING THE TEA LEAVES.
-
8/3/2019 2555571 Six Sigma and Minitab 13
143/396
Common Patterns of Variation
Within Piece
Characterized by largevariation in readings
taken of the same singlesample, often fromdifferent positions withinthe sample.
Piece to Piece
Characterized by largevariation in readings
taken between samplestaken within a short timeframe.
Time to Time
Characterized by largevariation in readings
taken between samplestaken in groups with asignificant amount oftime elapsed betweengroups.
MULTI-VARI EXERCISE
-
8/3/2019 2555571 Six Sigma and Minitab 13
144/396
We have a part dimension which is considered to be impossible to manufacture. A
capability study seems to confirm that the process is operating with a Cpk=0(500,000 ppm). You and your team decide to use a Multi-Vari chart to localize thepotential sources of variation. You have gathered the following data:
Sample Day/Time
Beginningof Part
Middle ofPart
End ofPart
1 1/0900 .015 .017 .0182 1/0905 .010 .012 .015
3 1/0910 .013 .015 .016
4 2/1250 .014 .015 .018
5 2/1255 .009 .012 .0176 2/1300 .012 .014 .016
7 3/1600 .013 .014 .017
8 3/1605 .010 .013 .015
9 3/1610 .011 .014 .0179
Construct a multi-vari chart of the data
and interpret theresults.
Learning Objectives
-
8/3/2019 2555571 Six Sigma and Minitab 13
145/396
Understand how to use multi-vari charts in completing animprovment project.
Know how to properly gather data to construct multi-vari
charts.
Know how to construct a multi-vari chart.
Know how to interpret a multi-vari chart.
-
8/3/2019 2555571 Six Sigma and Minitab 13
146/396
SAMPLE SIZE
CONSIDERATIONS
Learning Objectives
-
8/3/2019 2555571 Six Sigma and Minitab 13
147/396
Understand the critical role having the right sample sizehas on an analysis or study.
Know how to determine the correct sample size for a
specific study.
Understand the limitations of different data types onsample size.
Why do we Care?
-
8/3/2019 2555571 Six Sigma and Minitab 13
148/396
The correct sample size isnecessary to:
ensure any tests you designhave a high probability of
success.properly utilize the type of datayou have chosen or are limited toworking with.
IMPROVEMENT ROADMAPUses of Sample Size Considerations
-
8/3/2019 2555571 Six Sigma and Minitab 13
149/396
Uses of Sample Size Considerations
Breakthrough
Strategy
Characterization
Phase1:
Measurement
Phase 2:Analysis
Optimization
Phase 3:
Improvement
Phase4:
Control
Sample Sizeconsiderations are used inany situation where asample is being used to
infer a populationcharacteristic.
Common Uses
KEYS TO SUCCESS
-
8/3/2019 2555571 Six Sigma and Minitab 13
150/396
Use variable data wherever possible
Generally, more samples are better in any study
When there is any doubt, calculate the needed sample size
Use the provided excel spreadsheet to ease sample size calculations
CONFIDENCE INTERVALS
-
8/3/2019 2555571 Six Sigma and Minitab 13
151/396
The possibility of error exists in almost every system. This goes for point values aswell. While we report a specific value, that value only represents our best estimatefrom the data at hand. The best way to think about this is to use the form:
true value = point estimate +/- error
The error around the point value follows one of several common probability
distributions. As you have seen so far, we can increase our confidence is to gofurther and further out on the tails of this distribution.
PointValue
+/- 1s = 67% Confidence Band
+/- 2s = 95% Confidence Band
This error band
which exists around
the point estimate iscalled the confidenceinterval.
BUT WHAT IF I MAKE THE WRONG DECISION?
-
8/3/2019 2555571 Six Sigma and Minitab 13
152/396
Not different (Ho)
Reality
TestD
ecision
Different (H1)
Not different (Ho)
Different (H1)
Correct Conclusion
Correct ConclusionaRisk Type I Error Producer Risk
Type II Error risk Consumer Risk
Test Reality = Different
Decision Point
Risk aRisk
WHY DO WE CARE IF WE HAVE THE TRUE VALUE?How confident do you want to be that you have made the right decision?
-
8/3/2019 2555571 Six Sigma and Minitab 13
153/396
How confident do you want to be that you have made the right decision?
Ho: Patient is not sick
H1: Patient is sick
Error Impact
Type I Error = Treating a patient who is not sick
Type II Error = Not treating a sick patient
A person does not feel well and checks into a hospital for tests.
Not different (Ho)
Reality
TestDecision
Different (H1)
Not different (Ho)
Different (H1)
Correct Conclusion
Correct ConclusionaRisk Type I Error Producer Risk
Type II Error risk Consumer Risk
-
8/3/2019 2555571 Six Sigma and Minitab 13
154/396
CONFIDENCE INTERVAL FORMULAS
-
8/3/2019 2555571 Six Sigma and Minitab 13
155/396
These individual formulas are not critical at this point, but notice that the onlyopportunity for decreasing the error band (confidence interval) without decreasing theconfidence factor, is to increase the sample size.
X t t na n a n + -/ , / ,2 1 2 1s sn XMean
sn
sn
a a
- --
1 1
1 22
22c s c/ /
Standard Deviation
Cpn
Cp Cpn
a n a nc1 2 12 2 121 1
- - -- -/ , / ,Process Capability
( ) ( )$
$ $
$
$ $
/ /p Z
p p
n p p Z
p p
na a-
+-
2 2
1 1
Percent Defective
SAMPLE SIZE EQUATIONS
-
8/3/2019 2555571 Six Sigma and Minitab 13
156/396
n ZX
a=-
/ 2
2
s
ns a
= +
sc
2
22 1/
Allowable error = -
(also known as )
X
Allowable error = s/s
( )n p pZ
E= - $ $/1 2
2
a Allowable error = E
Standard Deviation
Mean
Percent Defective
SAMPLE SIZE EXAMPLE
-
8/3/2019 2555571 Six Sigma and Minitab 13
157/396
Calculation Values:
Average tells you to use the mean formula
Significance: a = 5% (95% confident)Za/2 = Z.025 = 1.96
s=10 pounds
-x = error allowed = 2 pounds
Calculation:
Answer: n=97 Samples
We want to estimate the true average weight for a part within 2 pounds.
Historically, the part weight has had a standard deviation of 10 pounds.We would like to be 95% confident in the results.
nZ
X= -
= =
a
s
/
. *2
2 21 96 10
2 97
SAMPLE SIZE EXAMPLE
-
8/3/2019 2555571 Six Sigma and Minitab 13
158/396
Calculation Values:
Percent defective tells you to use the percent defect formula
Significance: a = 5% (95% confident)Za/2 = Z.025 = 1.96
p = 10% = .1
E = 1% = .01
Calculation:
Answer: n=3458 Samples
We want to estimate the true percent defective for a part within 1%.Historically, the part percent defective has been 10%. We would like tobe 95% confident in the results.
( ) ( )n p p ZE
= - = -
=
$ $ . ..
./
1 1 1 11 96
013458
2
2 2
a
Learning Objectives
-
8/3/2019 2555571 Six Sigma and Minitab 13
159/396
Understand the critical role having the right sample sizehas on an analysis or study.
Know how to determine the correct sample size for a
specific study.
Understand the limitations of different data types onsample size.
-
8/3/2019 2555571 Six Sigma and Minitab 13
160/396
CONFIDENCE INTERVALS
Learning Objectives
-
8/3/2019 2555571 Six Sigma and Minitab 13
161/396
Understand the concept of the confidence interval andhow it impacts an analysis or study.
Know how to determine the confidence interval for a
specific point value.
Know how to use the confidence interval to test futurepoint values for significant change.
Why do we Care?
-
8/3/2019 2555571 Six Sigma and Minitab 13
162/396
Understanding the confidenceinterval is key to:
understanding the limitations ofquotes in point estimate data.
being able to quickly andefficiently screen a series of pointestimate data for significance.
IMPROVEMENT ROADMAPUses of Confidence Intervals
-
8/3/2019 2555571 Six Sigma and Minitab 13
163/396
Breakthrough
Strategy
Characterization
Phase1:
Measurement
Phase 2:
Analysis
Optimization
Phase 3:
Improvement
Phase4:
Control
Used in any situationwhere data is beingevaluated for significance.
Common Uses
KEYS TO SUCCESS
-
8/3/2019 2555571 Six Sigma and Minitab 13
164/396
Use variable data wherever possible
Generally, more samples are better (limited only by cost)
Recalculate confidence intervals frequently
Use an excel spreadsheet to ease calculations
-
8/3/2019 2555571 Six Sigma and Minitab 13
165/396
So, what does this do for me?
-
8/3/2019 2555571 Six Sigma and Minitab 13
166/396
The confidence intervalestablishes a way to test whetheror not a significant change hasoccurred in the sampledpopulation. This concept iscalled significance or hypothesistesting.
Being able to tell when asignificant change has occurredhelps in preventing us from
interpreting a significant changefrom a random event andresponding accordingly.
REMEMBER OUR OLD FRIEND SHIFT & DRIFT?
-
8/3/2019 2555571 Six Sigma and Minitab 13
167/396
All of our work was focused in a narrow time frame.Over time, other long term influences come and gowhich move the population and change some of itscharacteristics.
Confidence Intervals give usthe tool to allow us to be able tosort the significant changes fromthe insignificant.
Original Study
USING CONFIDENCE INTERVALS TO SCREEN DATA
-
8/3/2019 2555571 Six Sigma and Minitab 13
168/396
2 3 4 5 6 7
TIME
95% ConfidenceInterval
Significant Change?
WHAT KIND OF PROBLEM DO YOU HAVE?
Analysis for a significant change asks the question What happened tomake this significantly different from the rest?
Analysis for a series of random events focuses on the process and asksthe question What is designed into this process which causes it to have
this characteristic?.
CONFIDENCE INTERVAL FORMULAS
-
8/3/2019 2555571 Six Sigma and Minitab 13
169/396
These individual formulas enable us to calculate the confidence interval for many ofthe most common metrics.
Mean
sn
sn
a a
- --
1 1
1 22
22c s c/ /
Standard Deviation
Cpn
Cp Cpn
a n a nc1 2 12 2 121 1
- - -- -/ , / ,Process Capability
( ) ( )$ $ $ $ $ $/ /p Z p pn p p Z p pna a- + -2 21 1Percent Defective
X t tn
a n a n + -/ , / ,2 1 2 1n
Xs s
CONFIDENCE INTERVAL EXAMPLE
-
8/3/2019 2555571 Six Sigma and Minitab 13
170/396
Calculation Values:
Average defect rate of 14,000 ppm = 14,000/1,000,000 = .014
Significance: a = 5% (95% confident) Za/2 = Z.025 = 1.96
n=10,000
Comparison defect rate of 23,000 ppm = .023
Calculation:
Answer: Yes, .023 is significantly outside of the expected 95% confidence interval of.012 to .016.
Over the past 6 months, we have received 10,000 parts from a vendor with
an average defect rate of 14,000 dpm. The most recent batch of partsproved to have 23,000 dpm. Should we be concerned? We would like tobe 95% confident in the results.
( ) ( )$
$ $
$
$ $
/ /p Z
p p
n p p Z
p p
na a- + -2 2
1 1
( ) ( ). .
. .
,. .
. .
,014 1 96
014 1 014
10 000014 1 96
014 1 014
10 000
- + -p . . . .014 0023 014 0023 +p. .012 016p
CONFIDENCE INTERVAL EXERCISE
-
8/3/2019 2555571 Six Sigma and Minitab 13
171/396
We are tracking the gas mileage of our late model ford and find thathistorically, we have averaged 28 MPG. After a tune up at Billy Bobsauto repair we find that we only got 24 MPG average with a standarddeviation of 3 MPG in the next 16 fillups. Should we be concerned? Wewould like to be 95% confident in the results.
What do you think?
Learning Objectives
-
8/3/2019 2555571 Six Sigma and Minitab 13
172/396
Understand the concept of the confidence interval andhow it impacts an analysis or study.
Know how to determine the confidence interval for a
specific point value.
Know how to use the confidence interval to test futurepoint values for significant change.
-
8/3/2019 2555571 Six Sigma and Minitab 13
173/396
Learning Objectives
-
8/3/2019 2555571 Six Sigma and Minitab 13
174/396
Understand how to select the correct control chart for anapplication.
Know how to fill out and maintain a control chart.
Know how to interpret a control chart to determine theoccurrence of special causes of variation.
Why do we Care?
-
8/3/2019 2555571 Six Sigma and Minitab 13
175/396
Control charts are useful to:
determine the occurrence ofspecial cause situations.
Utilize the opportunities
presented by special causesituations to identify and correct
the occurrence of the special
causes .
IMPROVEMENT ROADMAPUses of Control Charts
-
8/3/2019 2555571 Six Sigma and Minitab 13
176/396
Breakthrough
Strategy
Characterization
Phase1:
Measurement
Phase 2:
Analysis
Optimization
Phase 3:
Improvement
Phase4:
Control
Control charts can beeffectively used todetermine special cause
situations in theMeasurement andAnalysis phases
Common Uses
-
8/3/2019 2555571 Six Sigma and Minitab 13
177/396
What is a Special Cause?
-
8/3/2019 2555571 Six Sigma and Minitab 13
178/396
Remember our earlier work with confidence intervals? Any occurrence which falls
outside the confidence interval has a low probability of occurring by random chanceand therefore is significantly different. If we can identify and correct the cause, we
have an opportunity to significantly improve the stability of the process. Due to theamount of data involved, control charts have historically used 99% confidence fordetermining the occurrence of these special causes
PointValueSpecial cause
occurrence.
X
+/- 3s = 99% Confidence Band
What is a Control Chart ?
-
8/3/2019 2555571 Six Sigma and Minitab 13
179/396
A control chart is simply a run chart with confidence intervals calculated and drawn in.These Statistical control limits form the trip wires which enable us to determine
when a process characteristic is operating under the influence of a Special cause.
+/- 3s =
99% ConfidenceInterval
So how do I construct a control chart?
-
8/3/2019 2555571 Six Sigma and Minitab 13
180/396
First things first:Select the metric to be evaluated
Select the right control chart for themetric
Gather enough data to calculate thecontrol limits
Plot the data on the chart
Draw the control limits (UCL & LCL)
onto the chart.Continue the run, investigating andcorrecting the cause of any out of
control occurrence.
How do I select the correct chart ?
-
8/3/2019 2555571 Six Sigma and Minitab 13
181/396
What type ofdata do I have?
Variable Attribute
Counting defects or
defectives?
X-s Chart IMR ChartX-R Chart
n > 10 1 < n < 10 n = 1Defectives Defects
What subgroup size
is available?
Constant
Sample Size?
Constant
Opportunity?
yes yesno no
np Chart u Chartp Chart c Chart
Note: A defective unit can have
more than one defect.
How do I calculate the control limits?
X R Chart-
-
8/3/2019 2555571 Six Sigma and Minitab 13
182/396
CL X=
UCL X RA= + 2LCL X RA= - 2
X = average of the subgroup averages
R = average of the subgroup range values
2A = a constant function of subgroup size (n)
For the averages chart:
CL R=
UCL RD= 4LCL RD= 3
For the range chart:
n D4 D3 A2
2 3.27 0 1.88
3 2.57 0 1.02
4 2.28 0 0.73
5 2.11 0 0.58
6 2.00 0 0.48
7 1.92 0.08 0.42
8 1.86 0.14 0.37
9 1.82 0.18 0.34
UCL = upper control limit
LCL = lower control limit
-
8/3/2019 2555571 Six Sigma and Minitab 13
183/396
How do I calculate the control limits?
c and u Charts
-
8/3/2019 2555571 Six Sigma and Minitab 13
184/396
UCL U U
nu = + 3
U= total number of nonconformities/total units evaluated
n = number evaluated in subgroup
For varied opportunity (u): For constant opportunity (c):
C= total number of nonconformities/total number of subgroups
Note: U charts have an individually calculated control limit for each point plotted
LCL U U
nu = - 3
UCL C C C = +3
LCL C CC = -3
How do I interpret the charts?
-
8/3/2019 2555571 Six Sigma and Minitab 13
185/396
The process is said to be out of control if: One or more points fall outside of the control limits
When you divide the chart into zones as shown and:
2 out of 3 points on the same side of the centerline in Zone A
4 out of 5 points on the same side of the centerline in Zone A or B
9 successive points on one side of the centerline
6 successive points successively increasing or decreasing 14 points successively alternating up and down
15 points in a row within Zone C (above and/or below centerline)
Zone A
Zone B
Zone C
Zone C
Zone B
Zone A
Upper Control Limit (UCL)
Lower Control Limit (LCL)
Centerline/Average
-
8/3/2019 2555571 Six Sigma and Minitab 13
186/396
Learning Objectives
-
8/3/2019 2555571 Six Sigma and Minitab 13
187/396
Understand how to select the correct control chart for anapplication.
Know how to fill out and maintain a control chart.
Know how to interpret a control chart to determine out ofcontrol situations.
-
8/3/2019 2555571 Six Sigma and Minitab 13
188/396
-
8/3/2019 2555571 Six Sigma and Minitab 13
189/396
-
8/3/2019 2555571 Six Sigma and Minitab 13
190/396
-
8/3/2019 2555571 Six Sigma and Minitab 13
191/396
-
8/3/2019 2555571 Six Sigma and Minitab 13
192/396
-
8/3/2019 2555571 Six Sigma and Minitab 13
193/396
-
8/3/2019 2555571 Six Sigma and Minitab 13
194/396
-
8/3/2019 2555571 Six Sigma and Minitab 13
195/396
HYPOTHESIS TESTING ROADMAP...
-
8/3/2019 2555571 Six Sigma and Minitab 13
196/396
Test Used
F Stat (n>30)
F Stat (n5)
Test Used
Z Stat (n>30)
Z Stat (p)
t Stat (n
-
8/3/2019 2555571 Six Sigma and Minitab 13
197/396
Determine the hypothesis to be tested (Ho:=, < or >). Determine whether this is a 1 tail (a) or 2 tail (a/2) test. Determine the a risk for the test (typically .05). Determine the appropriate test statistic.
Determine the critical value from the appropriate test statistic table.
Gather the data.
Use the data to calculate the actual test statistic.
Compare the calculated value with the critical value.
If the calculated value is larger than the critical value, reject the nullhypothesis with confidence of 1-a (ie there is little probability (p
-
8/3/2019 2555571 Six Sigma and Minitab 13
198/396
Many test statistics use a and others use a/2 and often it is confusing to
tell which one to use. The answer is straightforward when you considerwhat the test is trying to accomplish.
If there are two bounds (upper and lower), the a probability must be splitbetween each bound. This results in a test statistic of a/2.
If there is only one direction which is under consideration, the error(probability) is concentrated in that direction. This results in a teststatistic of a.
CriticalValue
RareOccurrence
CommonOccurrence
a Probability1-a Probability
CriticalValue
CriticalValue
CommonOccurrence Rare
OccurrenceRare
Occurrence
a/2 Probability1-a Probabilitya/2 Probability
Test Fails in either direction = a/2Test Fails in one direction = a
a a/2Examples
Ho: 1
>2
Ho: s1
-
8/3/2019 2555571 Six Sigma and Minitab 13
199/396
t
n n
n s n s
n n
=+ - + -
+ -1 212
2]2
1 2
1 1 1 1
2
[( ) ( )| |tdCALC X XR R=-
+2 1 2
1 2
ZX X
n n
CALC =-
+
1 2
12
1
212
2
s s
n n1 2 20= n n1 2 30+ >n n1 2 30+
2 Sample Tau 2 Sample Z2 Sample t
(DF: n1+n2-2)
Use these formulas to calculate the actual statistic for comparisonwith the critical (table) statistic. Note that the only major determinatehere is the sample sizes. It should make sense to utilize the simpler
tests (2 Sample Tau) wherever possible unless you have a statisticalsoftware package available or enjoy the challenge.
X X-1 2
Hypothesis Testing Example (2 Sample Tau)
Several changes were made to the sales organization The weekly
-
8/3/2019 2555571 Six Sigma and Minitab 13
200/396
Ho: 1 = 2
Statistic Summary: n1 = n2 = 5
Significance: a/2 = .025 (2 tail) taucrit = .613 (From the table for a = .025 & n=5)
Calculation: R1=337, R2= 577
X1=2868, X2=2896
tauCALC=2(2868-2896)/(337+577)=|.06|
Test: Ho: tauCALC< tauCRIT Ho: .06
-
8/3/2019 2555571 Six Sigma and Minitab 13
201/396
Ho: 1 = 2
Statistic Summary: n1 = n2 = 5
DF=n1 + n2 - 2 = 8
Significance: a/2 = .025 (2 tail) tcrit = 2.306 (From the table for a=.025 and 8 DF)
Calculation: s1=130, s2= 227
X1=2868, X2=2896
tCALC=(2868-2896)/.63*185=|.24|
Test: Ho: tCALC< tCRIT Ho: .24 < 2.306 = true? (yes, therefore we will fail to reject the null hypothesis).
Conclusion: Fail to reject the null hypothesis (ie. The data does not support the conclusion that there is a significantdifference
orders were tracked both before and after the changes. Determine if the sampleshave equal means with 95% confidence.
Receipts 1 Receipts 23067 32002730 2777
2840 26232913 30442789 2834
t
n n n s n sn n
=+ - + -+ -1 2 1
2
2]
2
1 21 1 1 12[( ) ( )
X X-1 2
-
8/3/2019 2555571 Six Sigma and Minitab 13
202/396
Sample Variance vs Sample Variance (s2)Coming up with the calculated statistic...
-
8/3/2019 2555571 Six Sigma and Minitab 13
203/396
Use these formulas to calculate the actual statistic for comparisonwith the critical (table) statistic. Note that the only major determinateagain here is the sample size. Here again, it should make sense toutilize the simpler test (Range Test) wherever possible unless youhave a statistical software package available.
F sscalc
=2
2 =FRRCALC
MAX n
MIN n
,
,
1
2
Range Test F Test(DF1: n1-1, DF2: n2-1)
n1 < 10, n2 < 10 n1 > 30, n2 > 30
MAX
MIN
-
8/3/2019 2555571 Six Sigma and Minitab 13
204/396
HYPOTHESIS TESTING, PERCENT DEFECTIVE
n > 30 n > 30
-
8/3/2019 2555571 Six Sigma and Minitab 13
205/396
Use these formulas to calculate the actual statistic for comparisonwith the critical (table) statistic. Note that in both cases the individualsamples should be greater than 30.
( )Z
p p
p p
n
CALC =-
-
1 0
0 01Z
p p
np n p
n n
n p n p
n n n n
CALC =-
+
- +
+
1 2
1 1 2 2
1 2
1 1 2 2
1 2 1 21
1 1_ _
Compare to target (p0) Compare two populations (p1 & p2)
n1 > 30, n2 > 30
How about a manufacturing example?
We have a process which we have determined has a critical
-
8/3/2019 2555