Cheat sheet
CSSC Green Belt Cheat Sheet
Six Sigma Foundations
- Not published of exam
- Belt Levels + Roles Related Quality Methods Problem + Quality Framing Project + Team Setup SIPOC
Lean Concepts
- Not published of exam
- Seven Muda + Extras Lean Tools + Process Waste Picker 5S Phases Type I vs Type II Muda
DMAIC + Project Tools
- Not published of exam
- DMAIC Phase Deliverables Analyze + Improve Tools DMAIC vs DMADV Repeatability vs Reproducibility FMEA RPN
Statistics + Measurement
- Not published of exam
- Defect + Yield Formulas Sigma Conversion Table Distribution Basics Hypothesis Testing Terms Hypothesis Test Picker Correlation + Regression
SPC + Capability
- Not published of exam
- Control Chart Picker Capability + Control Limits Out-of-Control Tests Control vs Capability Common vs Special Cause
Quick Facts
- Exam CSSC Green Belt
- Path Standard Exam
- Questions 100
- Types Multiple choice, true/false
- Time 2 hours
- Pass 280 of 400 points
- Percent 70%
- Books Open book
- Fee $195 USD
- Attempts 3 within 1 year
- Prerequisites None
- Project Not required
- Validity No expiration
- Blueprint 24 sections, no weights
SIPOC FO
- Suppliers Inputs Process Outputs Customers
- Drafted in Define One brainstorming session Scales to any process
Belt Levels + Roles FO
- Belt order White, Yellow, Green, Black, Master
- Green Belt Collects data under Black Belt
- Green Belt scope Keeps day job; smaller projects
- Black Belt Directs Green Belt analysis
- Master Black Belt Reviews analysis before presenting
- Sponsor / champion Senior leader owning the result Owns
- Champion duties Funds, resources, corporate politics
- Process owner Metrics, procedures, control plan
Related Quality Methods FO
- Six Sigma Reduces variation and defects
- Lean Removes waste (muda)
- Kaizen Change for the better
- TQM Early enterprise-wide quality program
- BPR Radical technical process redesign
- Rummler-Brache Nine Boxes performance model
- Nine Boxes grid Performer, process, organization levels
- JumpStart Fast fix, no statistics Low risk only
Problem + Quality Framing FO
- y = f(x) Output caused by inputs
- 5 Whys Drill past the symptom
- Problem statement Where, when, magnitude, metric
- Statement excludes No cause, no solution
- VOC Voice of the customer
- CTQ tree Need, drivers, measurable requirements
- CoPQ Cost of poor quality
- Internal failure Caught before delivery
- External failure Found by the customer
Project + Team Setup FO
- Project charter Output of Define phase
- Charter contents Problem, CTQs, roles, duration
- SIPOC Suppliers Inputs Process Outputs Customers
- Is / Is Not matrix Fixes scope boundaries early
- Stakeholder analysis Power versus interest grid
- Project viability model 15 weighted selection criteria
- Critical path method Dependencies set the schedule
- Tollgate review Sponsor approves phase exit
5S Phases LE
- Sort, Straighten, Shine, Standardize, Sustain
- Sort: remove clutter Straighten: a place for each Shine: clean and inspect Standardize: written standard Sustain: everyone commits
Type I vs Type II Muda LE
Type I
- Non-value-added
- Currently essential
- Make it efficient
Type II
- Non-value-added
- Not essential
- Remove immediately
- Streamline vs delete
Waste Picker LE
- Made too early → Overproduction
- Stock stacks before step → Inventory ( Bottleneck )
- Fixing defective output → Correction ( Rework )
- People walking needlessly → Motion
- Parts moved needlessly → Conveyance
- Idle between steps → Waiting
- Steps customer never values → Over-processing
- Skills going unused → Talent ( CSSC addition )
Seven Muda + Extras LE
- Overproduction Too much, too soon
- Correction Rework and defect fixing
- Inventory Stock piling before a step
- Motion Needless movement of people
- Conveyance Needless movement of material
- Over-processing More work than required
- Waiting Idle time between steps
- CSSC additions Talent, ideas, capital
- Type I muda Non-value-added but currently essential
- Type II muda Non-essential; remove immediately
Seven Muda LE
- Overproduce, Correct, Inventory, Motion, Convey, Process, Wait
- Plus talent Plus ideas Plus capital
Lean Tools + Process LE
- 5S Sort Straighten Shine Standardize Sustain
- Sustain Hardest phase; needs commitment
- Just-in-time Produce only what's needed
- Value stream map Exposes waste across process
- Bottleneck Slowest step caps output
- Smaller batches Cut lead time, inventory
- Process components Inputs outputs events tasks decisions
- Process definition layers Steps, time, dependencies, resources
- Poka yoke Mistake-proofs the process
DMAIC Order DM
- Define Measure Analyze Improve Control
- Define: charter Measure: baseline Analyze: root cause Improve: pilot Control: control plan
DMAIC vs DMADV DM
DMAIC
- Improves existing process
- Define Measure Analyze
- Improve then Control
DMADV
- Creates new process
- Define Measure Analyze
- Design then Verify
- Fix existing vs design new
DMAIC Phase Deliverables DM
- Define Charter, SIPOC, scope
- Measure Baseline data, FMEA, metrics
- Analyze Root causes, validated statistically
- Improve Select, pilot, implement
- Control Control plan, SPC, handoff
- Design (DMADV) Build the new process
- Verify (DMADV) Confirm the design performs
- Phase purpose Eat the elephant gradually
FMEA RPN DM
- RPN = Severity x Occurrence x Detection
- Score each 1-10 Highest RPN first Rescore after the fix
Repeatability vs Reproducibility DM
Repeatability
- One appraiser
- Same part, same gauge
- Same reading twice
Reproducibility
- Different appraisers
- Same part, same gauge
- Do they agree
- One person vs several people
Analyze + Improve Tools DM
- Fishbone bones People Process Materials Procedure
- Optional bones Equipment and Environment
- Pareto principle 20% causes, 80% effects
- FMEA RPN Severity x Occurrence x Detection
- Solutions selection matrix Ranks fixes against causes
- Cost benefit analysis Weighs gain against spend
- Pilot Limited live trial first
- Control plan Who monitors what, when
- Visual management Status visible at a glance
- Gage R&R Tests the measurement system
Type I vs Type II ST
Type I (alpha)
- Reject a true null
- False alarm
- Set by alpha
Type II (beta)
- Keep a false null
- Missed difference
- Power = 1 - beta
- False alarm vs missed signal
Hypothesis Test Picker ST
- One proportion versus target → 1-Proportion test ( Discrete data )
- Two proportions compared → 2-Proportion test ( Discrete data )
- Mean versus target, normal → 1-Sample t test ( Small sample )
- Same group before, after → Paired t test ( Same x factor )
- Means of two populations → 2-Sample t test ( Different x factors )
- Comparing variance or deviation → Chi-Square test ( Also 1-Variance test )
- Median versus target, non-normal → 1-Sample Wilcoxon ( Fairly symmetrical data )
- Non-normal and badly skewed → 1-Sample Sign test ( Wilcoxon alternative )
- Two medians, non-normal → Mann-Whitney test ( Two x factors )
Defect + Yield Formulas ST
- DPMO (defects / opportunities) x 1,000,000
- DPMO example (2 / 900) x 1,000,000 = 2,222
- DPU defects / units sampled
- DPU example 9 / 50 = 0.18
- Yield (opportunities - defects) / opportunities
- FTY good units / units entering
- RTY (entering - scrap - rework) / entering
- Multi-step yield Multiply each step's yield
- Why RTY is lower It charges rework as loss
Discrete vs Continuous Data ST
Discrete
- Nominal, ordinal, binary
- Counted categories
- Needs larger samples
Continuous
- Measured in units
- More precise
- Collect this where possible
- Continuous converts down, not up
Distribution Picker ST
- Pass or fail, fixed trials → Binomial ( Discrete )
- Metric reads 'per unit' → Poisson ( Discrete counts )
- Trials until first defect → Geometric ( Waiting time )
- Time between failures → Exponential ( Continuous )
- Skewed durations above zero → Lognormal ( Continuous )
- Reliability, shape varies → Weibull ( Distribution family )
- Symmetrical continuous data → Normal ( Still test normality )
Sigma Conversion Table ST
- 99.7450% yield 2,550 DPMO = 4.3 sigma
- 99.6540% yield 3,460 DPMO = 4.2 sigma
- 99.5340% yield 4,550 DPMO = 4.1 sigma
- 99.3790% yield 6,210 DPMO = 4.0 sigma
- 99.1810% yield 8,190 DPMO = 3.9 sigma
- Six sigma 3.4 DPMO, 99.99966% yield Goal
- Worked example 99.5% yield = 4.0-4.1 sigma
FTY vs RTY ST
FTY
- Good units / units in
- Ignores rework
- Looks better
RTY
- Subtracts scrap and rework
- Probability of defect-free unit
- Always lower
- Rework hidden vs rework charged
Distribution Basics ST
- Normal Symmetrical continuous bell curve
- Binomial Two outcomes, independent trials
- Poisson Counts per time or space
- Poisson clue The word 'per' in metric
- Geometric Trials before first occurrence
- Exponential Arrival times, between failures
- Lognormal Right-skewed durations above zero
- Weibull Family imitating other shapes
- Logistic Approximates the normal curve
- Gamma Always skewed to right
Hypothesis Testing Terms ST
- H0 Null: no difference exists
- Ha Alternative: a difference exists
- p-value rule p below alpha, reject H0
- Fail to reject Not proof H0 true
- Alpha Type I error risk
- Beta Type II error risk
- Power 1 - beta
- Beta 0.20 Power of 0.80
- Delta Smallest difference worth detecting
- Sample size inputs Alpha, beta, delta
- Random sampling Required, or inference fails
Correlation + Regression ST
- Scatter diagram Shows relationship, never causation
- r range -1 through 1
- r near 0 No relationship at all
- r at 1 Every point on line
- Pearson Usual correlation calculation
- Regression Predicts y from x
- Chi-squared goodness-of-fit Tests data for normality
- Beyond 3 sigma Under 1% of data
Chart Data Split SP
- Variable: X-bar, I-MR. Attribute: p, np, u, c
- p, np: defective units u, c: defect counts np, c: constant size
Control vs Capability SP
In control
- Low variation
- Control limits from data
- Says nothing about spec
Capable
- Low variation too
- Centered on requirement
- Measured by Cpk
- Steady vs steady and right
Control Chart Picker SP
- Continuous, subgroups under 8 → X-bar & R ( Mean plus range )
- Continuous, subgroups over 8 → X-bar & S ( Sigma easily calculated )
- Continuous, no sensible subgroups → I-MR ( Individual points )
- Data slow or costly → I-MR ( Cannot wait for subgroups )
- Discrete, percent defective → p chart ( Sample size varies )
- Discrete, percent, constant size → np chart ( Counts nonconforming units )
- Defect counts, size varies → u chart ( Defects per unit )
- Defect counts, constant size → c chart ( Defects per sample )
Capability + Control Limits SP
- Sigma level Center to nearest specification limit
- Sigma level formula (USL - median) / sigma
- Take which value The smaller of two
- Cpk sigma level / 3
- Cpk 1.33 Equals sigma level 4
- Cpk 2.0 Equals sigma level 6
- UCL 3 sigma above center
- LCL 3 sigma below center
- Zone C Nearest the center line
- Zone A Outermost, beside control limits
Common vs Special Cause SP
Common cause
- Built into process
- Expected variation
- Fix the system
Special cause
- Outside normal expectation
- Bad headset, new hire
- Investigate the event
- Process noise vs outside event
Out-of-Control Tests SP
- Test 1 One point beyond limits Act now
- Test 2 Nine in a row, one side
- Test 3 Six points trending steadily
- Test 4 Fourteen points alternating direction
- Test 5 Two of three in A
- Test 6 Four of five in B
- Test 7 Fifteen in a row, C
- Test 8 Eight in a row, skipping C
Common Traps
Sigma level vs priority
- Sigma flags worst variation ≠ Cost sets the priority
FTY vs RTY
- FTY hides rework ≠ RTY charges rework
Control vs spec limits
- Control limits from data ≠ Spec limits from customer
Correlation vs causation
- Scatter shows relationship ≠ It never proves cause
Failing to reject
- Not proof H0 true ≠ Only insufficient evidence
p charts vs u charts
- p tracks defective units ≠ u tracks defect counts
DMAIC vs DMADV
- DMAIC improves existing ≠ DMADV designs new
Alpha vs beta
- Alpha rejects true null ≠ Beta keeps false null
Cpk vs sigma level
- Cpk is sigma / 3 ≠ Cpk 1.33 equals 4
Last Minute
- 1. DPMO = defects / opportunities x 1,000,000
- 2. RTY subtracts scrap and rework
- 3. Multiply step yields for overall
- 4. Cpk = sigma level / 3
- 5. Cpk 1.33 equals sigma level 4
- 6. Control limits sit 3 sigma out
- 7. Subgroups under 8: X-bar & R
- 8. Subgroups over 8: X-bar & S
- 9. No subgroups: use I-MR chart
- 10. p and np track defective units
- 11. u and c track defect counts
- 12. np and c need constant size
- 13. p below alpha means reject H0
- 14. Power = 1 - beta
- 15. Seven muda plus talent, ideas, capital
- 16. 5S ends with Sustain
- 17. Charter is the Define output
- 18. Fishbone: People Process Materials Procedure
- 19. Exam: 100 questions, 2 hours
- 20. Pass needs 280 of 400
- 21. Open book: bring your manual
CSSC Certified Six Sigma Green Belt
- Practice Questions 125 questions Study Guide 54 topics Cheat Sheet 133 facts Flashcards 50 cards Shorts 1 short Free CE Podcast