DMAIC for BB Flashcards
(135 cards)
1
Q
What is the goal of Lean?
A
To reduce waste
2
Q
What is the goal of Six Sigma?
A
To reduce variation
3
Q
Are we focused on the Process or the Output?
A
We are focused on the Output of the Process.
4
Q
For Lean Six Sigma, what is the Problem that we are focused on?
A
The Output
5
Q
What do you find WITHIN the Process?
A
Root causes and solutions
The Problem is always the Output of the Process
6
Q
What does CTC/CTB stand for?
A
Critical to Customer/Critical to Business
7
Q
What does a good CTC statement start with?
A
Each or Every
8
Q
What does RACI stand for?
A
Responsible, Accountable, Consult, Inform
9
Q
What should your sample size be for Discrete data?
A
100 samples
10
Q
What should your sample size be for Continuous data?
A
30 samples
11
Q
What is Tool 2 used for?
A
The measurements of the OUTPUT on your CTCs.
How will you measure your OUTPUT?
12
Q
What makes a measurement strong for Tool 2?
A
A strong measurement is Output related and is something that could be measured by you or the customer at the end of the process
13
Q
Which capability Key Figures are used for BAD outputs?
A
DPMO & DPU & PPM
14
Q
How do we calculate Yield?
A
(# of good parts / # of total parts) x 100%
15
Q
How do we calculate PPM?
A
(# of Defect Parts / # of Total Parts) x 1,000,000
16
Q
Which Capability Key Figure is used for the Business perspective for a single part that could have multiple defects?
A
DPMO
17
Q
How do you calculate DPMO?
A
DINO
(# of Defects / (# of total parts x opportunities)) x 1,000,000
18
Q
What does RTY stand for?
A
Rolled Throughput Yield
19
Q
How do you calculate RTY?
A
Multiply the Yield percentages:
50% Yield in Time
90% Yield in Location
80% Yield in Star
0.50.90.8 =0.36
So 36%
20
Q
What does DPU stand for?
A
Defects per Unit
21
Q
How do you calculate DPU?
A
of Defects / # of Total Parts
Ex:
We produce 10 sunglasses that each have 4 opportunities for a Defect (CTC).
4/10
22
Q
What does Cp stand for?
A
Capability Probability
23
Q
What do USL AND LSL stand for?
A
Upper Specification Limit
Lower Specification Limit
24
Q
What does Cpk check for?
A
The Location of the data (is it in the middle or skewed right or left)
25
During which phases do we look at the As Is process?
Define, Measure, Analyze
26
For Change Management, what do we need to design & address constantly throughout the Transition Phase?
Communication Process
Attitudes
Mood Behaviour
27
What are the 6 phases of the Change Curve (Kubler Ross)?
Shock
Denial
Resistance
Test
Acceptance
Integration
28
What is the Kano Model used for?
The degree of realization vs the customer’s happiness. The better I fulfill, the happier the customer.
29
What are the 3 types of lines in the Kano Model?
Delighter (factors which enthuse)
Satisfier (Performance factor)
Dissatisfier (Basic Factors)
30
Which of the following should you address first according to the Kano Model?
Delighter, Satisfier, or Dissatisfier?
Dissatisfier as it is considered a basic factor/requirement of the customer.
31
What are we looking to find in MSA 1?
Variation in the process - the spread
32
What are the 2 main goals of the MSA 1?
Minimize influencing factors (sources of variation)
Ensure capability of Gage (requirements to our Gage)
33
What are the 6 sources of Variation?
Man
Machine
Method
Material
Measurement
Mother Nature
34
What 6 requirements of our Gage do we look into?
Accuracy
Stability
Linearity
Granularity
Repeatability
Reproducibility
35
What is Accuracy measuring?
How accurate our data is with the Mean. To the spot - to the point
36
What is Stability measuring?
The data over time. Is the process stable yesterday, today, and tomorrow? We want the same results over time.
37
What is the minimum # of Categories required for Granularity?
5 Categories
38
What are we measuring with Granularity?
Within the check of my measurement system, we need to be able to scale down to the granular level as necessary.
39
What do we check with Linearity?
Up/down scaling. Within the range of interest, your measurement scale must work.
Ex: You would not use a food scale to measure the weight of a fully grown shark
40
What is MSA 1 strong on?
Accuracy and Capability
41
What is Cg?
Capability of Gage
42
What are we measuring in MSA Type 2?
Repeatability & Reproducibility
43
What are the good and bad percentages for Contribution?
Good < 1%
Bad > 9%
44
What are the good and bad percentages for Study Variation?
Good < 10%
Bad > 30%
45
What are the Good and Bad percentages for Tolerance?
Good < 10%
Bad > 30%
46
How many NDC (Number of Distinct Categories) are we looking for?
More than or equal to 5
This is the same as Granularity
47
Contribution is…?
Variance Components
48
Study Variation is…?
The Gage Evaluation
49
Capability is…?
Is the As Is capable to meet the customer’s needs?
50
MSA is…?
Is the data I have something I can actually trust? Are we collecting in the right way? Consistently? Etc.
51
Which graphs do we use for a snapshot in time of Discrete data?
Bar Chart (categories)
Pie Chart
Pareto
52
What graphs do we use for a snapshot in time of Continuous data?
Histogram (sequence goes up and count is along the bottom)
Box Plot (single for 1 result - use multiple for comparison)
53
For DMAIC, you need a process that is…
Stable but not Capable
54
Why do we use Data Stratification?
To check the effect of an influencing factor (x) on the result (y)
55
In Analyze, what does it mean to use Passive data?
Passive data is using existing data. You do not need to collect it as it already exists
56
In Analyze, what does it mean to use Active data?
Active data is generating brand new data when there is no existing data available
57
What are the 4 steps in the cycle of Statistical Thinking?
Real Problem < Statistical Problem < Statistical Solution < Real Solution < Real Problem (back to the beginning)
58
What is the ‘mu’?
Mu is the mean of the Total Population
59
What is Granularity ‘typically’ calculated as in relation to your Standard Deviation?
Granularity is ‘typically’ 1/10th of your Standard Deviation (S)
60
DPMO is the performance indicator most representing the business perspective because…?
It includes all of the Opportunities for defects. It is not whole defective parts but the # of Defect Opportunities per part
61
When calculating the DPMO, what are the “opportunities for a defect” referring to?
Your CTCs and CTBs
62
Which 2 measurements count Good/Bad WHOLE parts?
Yield for Good Parts
PPM for Bad Parts
63
What is Ho?
The Null Hypothesis, which is always equal. The different cannot be detected.
64
What is Ha?
The Alternative Hypothesis. It is always Not Equal. This means that significant difference IS visible in my sample
65
If your data is not robust enough/good enough, which Hypothesis will you always apply?
Ho - Null Hypothesis
66
Finish the sentence:
If P is low…
Ho must go!!
67
For Ho are we testing if xbars are equal or xmus are equal?
xmu because we care about the Total Population
68
Where/How do I find mu?
You need a Sample and a Confidence Interval
69
If you are not given a Granularity, what should you use as your number?
1/10th the value of S
If your S is 12 then your Granularity could be assumed to be 1.2
70
If I increase my Confidence Level from 95%-99%, what happens to the Confidence INTERVAL if nothing else is changed?
It will INCREASE because my Z is increasing
71
What is the Alpha Risk?
The maximum risk I’m willing to take to refuse Ho (null hypothesis) wrongly
So if I have a 95% Confidence Interval, there’s a 5% chance I could be wrong and I’m willing to take that risk.
72
Who sets/decides the Alpha Risk?
Us/the producer because we decide our Confidence Interval
73
What does DOE stand for?
Design of experiment
74
Is a DOE Passive or Active data?
Active data as you need to generate new data
75
Which is harder to fix - Location or Spread?
Spread is MUCH harder to fix. Location will often 1-2 factors to fix whereas Spread could take 10
76
Which should you fix first: Location or Spread?
Spread because once you have very little spread it will be very easy to shift it to the desired Location
77
If your data is NOT Normal, what can you do?
Increase your sample size
Use Data Stratification
Look at statistical key figures
Live with it!
78
If your Mean and Median are close together, do you likely have Normal or Not Normal data?
Normal
79
What does ANOVA stand for?
ANalysis Of VAriance
80
What is the Basic Question asked by ANOVA?
How much of the Continuous observed distance (Y) is explained by one Discrete factor (X)?
81
True or False
A Factor in ANOVA can have more than one Factor Level
True
82
Does ANOVA test Spread or Location?
ANOVA tests Location
83
What does a Y with a line over it represent in ANOVA?
Expected Value
84
What does Tao (S with a line over it) represent in ANOVA?
Explained Variation
85
What does Epsilon (E) represent in ANOVA?
Unexplained Variation - Residual
86
In a Main Effect Plot, what does a Sloped Line and a Flat Line mean?
A Sloped Line indicates an effect. The higher the degree of slope, the stronger the effect.
A Flat Line indicates no effect.
87
How do you determine if there is an Interaction in an Interaction Plot?
If the lines intersect at any point in the provided data, there is an Interaction.
If the lines do NOT intersect at any point in the PROVIDED data, there is no interaction.
88
Should you extrapolate the lines in an Interaction Plot to find the Interaction point?
Do NOT extrapolate the lines! Only make the decision based on the provided data that you can visually see.
89
What is R^2 used for?
Explained variation in the model
90
What is a good and bad R^2 value?
Good R^2 > 80%
Bad R^2 < 80%
91
Do we expect Unexplained Variation to behave Normally or Not Normally?
We expect it to behave Normally
92
What does AD stand for?
Anderson-Darling
93
What is the Anderson-Darling test used for?
It’s a test for Normality that works the exact same was as P-Value. Below 0.05 is Not Normal
94
If all the Residual data is Normal, what does that tell you about your Model?
Your Model is Statistically Valid
(basically, we can trust it)
95
What are we checking with Correlation?
Is there a Linear Dependency of 2 Continuous Variables?
96
What are we checking with Regression?
Is there a Mathematical Model that describes the Linear Dependency of 2 Continuous Variables?
97
What type of graph do we use for Correlation?
Scatterplot
98
What letter do we use for the Pearson Correlation Coefficient?
r
99
What are the different r values and their meanings?
r ~ 1 : Perfect. Used when going from bottom left to upper right in a diagonal
r ~ -1 : Perfect. Used when going from upper left to bottom right in a diagonal
r ~ 0.8 : Strong. They’re clustered together, but not perfect.
0.08 > r > 0.05 : Medium. Data is more spread out but still trending in the same direction
r ~ 0 : No Correlation. Either randomly scattered points or a strange shape (not a straight line)
100
Can you use your Output (Y) as one of the factors?
Yes
101
What is the Basic Question we asked with Linear Regression?
Can we predict a desired Output?
102
If you can make a prediction with your Linear Regression, it means that Ho…
…must go! It IS significant!
103
A successful Regression allows us to stop what?
Trial and Error because it is no longer necessary since we can calculate exactly what we need
104
What is a Residual?
The difference between a fitted value and the actual value
105
In MSA Type 1, what do we check for?
Bias. It’s checking the Bias of the device/machine
106
Is MSA Type 1 data always Discrete or Continuous?
Continuous as we are measuring a device/machine
107
What does a Chi squared test look at?
Differences between more than 2 discrete factor levels
108
MSA is checking is the variation comes from…?
Part to Part
109
In Hypothesis Testing, the P-Value is the chance that…?
There is a mistake in rejecting Ho (null)
110
What is the difference between Cp and Cg?
Cp is about the Process and Cg is about the Gage
111
What is Ao on a graph?
The Intercept with Y
112
What is the Basic Question asked by Multiple Linear Regression?
Can we predict a desired output with 2 or more Independent Continuous Factors?
113
What does VIF stand for?
Variance Inflation Factor
114
What do we use VIF to check for?
We are checking the Independence of individual factors where compared with others (2+ factors in a model)
115
What VIF value indicates a Factor is Dependent on other factors?
VIF > 5
116
What do you do if you have multiple Factors with a VIF above 5?
Remove the factor with the highest VIF and rerun the analysis. Continue to do this until all remaining Factors have a VIF < 5.
117
When checking your Multiple Linear Regression results, in what order do you check them?
Check P-value
Check R^2
Check Residuals to determine if statistically valid
Check VIF
118
When do you use Active Data Analysis?
When you don’t have any Passive (existing) Data or if your Passive Data doesn’t show you anything of value
119
What are the types of planned/designed Active Data experiments in order from simplest to most complex?
Simple Comparison
One Factor at a Time (OFAT)
Two Way ANOVA
Regression (ANOVA for Discrete Data and Regression for Continuous Data)
Design of Experiment (DOE)
120
121
What should you check (in regards to the constraints of your project) when deciding what type of experiments to undertake for Active Data Analysis?
Time
Budget
Resource Availability
122
What are the Advantages and Obstacles of OFAT?
Advantages - easy to understand, well accepted
Obstacles - it’s trial & error, no visible interactions, high effort/time
123
When do we decide to use DOE?
During Analyze - to find root causes for your problem when there isn’t any available data that can do so. Use this to find relevant Factors as root causes for your problem
During Improve - find optimum settings for the desired goal
124
Explain what the numbers mean in the Fullfactorial 2^3
2 is the number of Factor Levels
3 is the number of Factors
125
What are the 2 types of Influencing Factors?
Steering Variables
Disturbing Variables
126
What are the 2 components of Steering Variables?
Factor (min/max)
Constant (document value)
127
What are the 2 components of Disturbing Variables
Known
Unknown - address through repetition and randomization
128
True or False
Interaction is not Correlation
True. They can Interact and still be Independent from each other
129
What are the components of a Fractional Factorial?
2^(k-q) IV
2 = Factor Level
k = # of Factors
q = Reduction (used to reduce trial numbers)
IV = Resolution Type
130
What is the cost/downside of using a Fractional Factorial?
Loss of information - blurriness in our data
131
Do Basic Trials include Trial Repeat?
No, they do not
132
What 3 things is Hypothesis Testing looking at during analysis, depending on if you have discrete or continuous data?
Proportion, Location (mu), and Spread (sigma^2)
133
What are we looking for with ANOVA?
Main Effect, Interaction Plot, Significance (P-value), R^2, & Residuals
134
How do you deal with noise in your data?
Randomization & Blocking (day shift vs night shift, etc)
135
In Improve, how do we generate solutions? Rank them in order
Obvious, Best Practice, Lean Toolbox, Creativity Techniques
