Please enable JavaScript.

Coggle requires JavaScript to display documents.

six sigma foundations (DMAIC (define (project team is selected (determine…

- - - - understand what is important to customers
      - determine performance outcome to be improved
        
        y = f(x)
  - - - process steps are understood
      - also consider time taken, delays, rework, etc.
    - - collect data
      - measure how long takes, consistency, etc.
  - - - perform root cause analysis
      - 5 whys
  - - - developed
      - piloted
      - implemented
  - - - ensure
        
        improvements or gains are sustainable
      - controls put in place
        
        employees know to intervene
        
        how
        
        when
- - - - quality concepts
      - process thinking
  - - - variation
      - voice of customer analysis
      - process mapping
      - root cause analysis
  - - - apply what they learn between the training sessions
    - - usually on part-time basis
  - - - broader scope
      - require more advanced statistical analysis
    - - advanced quality and statistical tools
        
        process capability
        
        inferential statistics
        
        design of experiments
  - - - advanced statistical analysis
      - process simulation
      - project management
      - change management
- - - - what measureable performance outcome must be accomplished? by when?
      - i.e. defines the Y in Y = f(x)
      - problem statement
        
        must be
        
        specific
        
        measureable
        
        in operational terms
        
        in financial terms
        
        relevant
        
        significant to the organization
        
        recurring
        
        example
        
        Over the past 12 months, First Call Resolution "at our IT Help Desk is only 60%. "This is below the 75% required "in our service level agreement" or SLA. "Failure to meet this requirement will result "in a loss of $200,000 in penalties, "not to mention customer dissatisfaction, "nonrenewals, and the potential loss of clients.
      - goal statement
        
        establish the target results to be achieved within a specific timeframe
        
        should be SMART
        
        specific
        
        what needs to be improved
        
        measureable
        
        how much improvement to achieve
        
        attainable
        
        target result is achievable
        
        relevant
        
        aligned with the success of the business
        
        time-bound
        
        timeframe for completion
        
        e.g. Improve First Call Resolution rate to 75% or higher, "while ensuring customer satisfaction, "within the next four months.
    - - contains
        
        project scope
        
        expected benefits
        
        operational
        
        financial
        
        speak the language of management, which is money
        
        project team names
        
        key roles
        
        milestones
        
        problem / opportunity statement
        
        project name
        
        goal statement
        
        key metric
        
        this is the Y in Y = f(x)
        
        signatures
        
        project leader
        
        finance
        
        project champion
      - serves as
        
        internal contract between senior management and project team
        
        common understanding on purpose and scope of project
      - reviewed and approved by project champion
      - purpose
        
        provides direction and focus to project team
        
        management authorization for the project
      - asd
        
        purpose
        
        goals
        
        scope of work
        
        assignment of resources
    - - SIPOC
        
        input
        
        process
        
        output
        
        customer
        
        purpose
        
        supplier
      - purpose
        
        documents and communicates overall scope of the processes related to the project, on one page
        
        is high-level, bird's eye view
  - - - measure the size and scope of the problem or performance gap
        
        how big
        
        how bad
        
        how widespread
      - measure the Y in Y = f(x)
    - - 1) develop data collection plan
        
        why? because collecting data is
        
        costly
        
        time consuming
        
        resource draining
        
        tips
        
        don't collect data on everything in sight
        
        instead plan your collection correctly
        
        collect only data needed to obtain information and answers to questions we want answered
        
        steps
        
        1) list specific questions to answer regarding Y
        2) decide how the answers will be presented and displayed
        3) select the tools that are needed to make that possible
        4) determine what type of data how much data are required by that tool
        5) specify where and from whom the data should be collected
      - 2) map current processes
        
        involve people who know the process
        
        decide on the start and end points
        
        use a logical orientation, left-to-right, top-to-bottom
        
        walk yourself through the process to understand it, what's happening to you
        
        avoid at this time
        
        analyzing
        
        improving
        
        have others review the final process map
      - 3) validate the measurement system
        
        where they measured correctly? recorded correctly?
        
        measurement system analysis (MSA)
        
        looks at
        
        operator
        
        procedures used to collect
        
        instrument
        
        how to conduct
        
        20-25 known items are selected blindly
        
        items are measured 2x by two operators
        
        do before any data is collected or used
        
        valid means
        
        accurate with no bias
        
        repeatable
        
        e.g. should be able to take more than one reading
        
        reproduceable
        
        e.g. weight at home should be same as in doctors office
      - 4) measure the Y in Y = f(x)
        
        how bad the problem is compared to requirements, specifications or tagets
        
        how well customer expectations are met
        
        how often are customer expectations met or not met?
        
        how widespread is it?
        
        process capability and sigma leve
        
        measure of how well a process performs against its target specifications
    - - data types
        
        continuous
        
        can occupy any value over a continuous range
        
        decimals, fractions
        
        discrete
        
        can only take on certain values
        
        integers, whole numbers
        
        number of departures
        
        aka categorical data
      - graphs
        
        best one is dicated by data type
        
        common graphs for continuous data
        
        histograms
        
        dotplots
        
        boxplots
        
        shows the mean and variation
        
        common graphs for discrete data
        
        bar charts
        
        pareto charts
        
        contrast the frequency count in each category
        
        trends over time
        
        line graphs
        
        time-series chart
        
        control charts
  - - - answer "why" there is a problem
      - determine which Xs impact Y in y = f(x)
    - - 1) generate a potential list of Xs that impact Y
        
        cause - effect diagram
        
        fish diagram
        
        tree diagram
      - 2) organize the potential Xs
        
        cause-effect (fishbone) diagrams
      - 3) shortlist and select the likely key Xx
      - 4) develop a data collection plan for the analysis
        
        what is the hypothesis to be tested?
        
        what type of hypothesis test should be used?
        
        what data is needed?
        
        how much data is needed?
        
        from whom, when, where?
        
        recommend laying this out in a table, each question as a column
      - 5) prove the key Xs in Y = f(x)
        
        inferential statistics
        
        hypothesis tests
        
        theory is presented
        
        data is gathered to prove / disprove the theory
        
        aka scientific method
    - - 1) identify any bottlenecks
        
        delays
        
        holdups
        
        why
      - 2) identify any workaorunds
        
        why so many steps
        
        examine each step, ask why needed, what value it adds
        
        VA value adding
        
        NVA non value adding
        
        VE value enabling
      - 3) zero in on the decision diamonds
        
        inspection
        
        checks
        
        evaluation decisions
        
        how often do they fail, why?
      - 4) look at rework loops
        
        how long are they?
      - 5) shorten the rework loop (number of reworked steps)
  - - - address proven Xs
      - find solution to improve Y
    - - 1) generate potential solutions to address proven key Xs
        
        brainstorming
        
        anti-brainstorming
        
        how to make things worse
        
        creative-thinking techniques
        
        Six Thinking Hats (Edward do Bono)
        
        optimism
        
        devil's advocate
        
        creativity
        
        process
        
        facts
        
        feelings
      - 2) evaluate solution alternatives
        
        multi-voting
        
        2x2 matrices
        
        e.g. high/low cost vs benefit
      - 3) select the right set of solutions and implement
        
        do before implementing
        
        failure modes and effects analysis (FMEA)
        
        types
        
        design FMEA
        
        process FMEA
        
        possilbe failure modes (of each step)
        
        2 more items...
        
        severity score (of each failure mode)
        
        2 more items...
        
        occurrence score (of each failure mode)
        
        2 more items...
        
        detection score (of each failure mode)
        
        2 more items...
        
        Risk Priority Number (RPN) = Severity x Occurrence x Detection
        
        highest failure modes are first target for improvements
        
        widely used in Six Sigma
        
        mistake proofing
        
        why
        
        prevents errors from occurring
        
        minimizes errors
        
        detects errors immediately
        
        levels (in order of preference)
        
        1) prevention
        
        e.g. resequence steps
        
        2) facilitation
        
        facilitation of work to minimze errors
        
        3) detection
        
        e.g. underline errors in auto-spellcheck
        
        pilot testing
        
        small scale test to ensure improvements will work as planned
        
        fix any problems, kinks before rolled out
        
        process maps of improved process
        
        selecting techniques
        
        criteria selection matrix
        
        evaluates and scores each solution alternative against agreed criteria
        
        weighted scores
        
        pugh matrix
        
        compare alternative solutions against baseline solution
        
        better, worse, same
        
        cost-benefit matrix
        
        implementation planning
        
        communication to ensure buy-in from key stakeholders
        
        procedures to be revised
        
        responsiblities and authority to change
        
        training to conduct
        
        timing of the implementation
        
        time and effort required for implementation
        
        budgeting for requires required
    - - involve key stakeholders
        
        process owners
        
        operators
        
        managers
        
        others impacted by problem
  - - - ensure improved Y is sustainable
    - - 1) develop a control plan
        
        monitoring, controlling, regulating
        
        when to leave process along
        
        when to take action, what action to take
      - 2) work with process owners to update procedures
      - 3) implement and monitor performance
      - 4) validate actual financial impact
      - 5) secure project completion sign-off
    - - indicates
        
        when to leave process alone
        
        when to intervene
        
        so you don't react to random/inherent variation
      - statistical process control
    - - 1) what needs to be controlled
        
        aka the control subject
        
        the relevant Y
        
        any key X factors
      - 2) specification, target, or desired target (or range)
      - 3) how feedback on actual performace is provided
      - 4) when to take action and what action to take
      - 5) who is reponsible and authorized to take action
- - - - types
        
        implementation projects
        
        solution is obvious
        
        just need to execute
        
        lean projects
        
        used when drivers of performance are easy to identify
        
        aka kaizen events or workouts
        
        six sigma (DMAIC) projects
        
        used when drivers of performance are unknown or must be analyzed
        
        design for six sigma (DMADV)
        
        used when new products, services, and processes are required
        
        used when existing products, services, and processes so borken that must be redesigned
      - use right methodology for type of project
    - - 1) identify and list customers
      - 2) group into segments based on needs, location or use
      - ways
        
        surveys
        
        but low response rates
        
        focus groups
        
        interviews
        
        can be skewed
        
        people who choose to participate are ones most disappointed
        
        observations
  - - - high-level
        
        usually 10 or fewer steps
        
        tells you how everything connets
      - detailed process
      - swimlane
        
        shows who does, where handoffs are
        
        handoffs are weakest links
    - - avoid
        
        judgement sampling
        
        told to use your judgement to select random, representative sample
        
        convenience sampling
        
        that are easiest for you to measure
        
        do use
        
        population sampling
        
        simple random sampling
        
        every item has the same odds of being selected
        
        stratified random sampling
        
        every group is represented proportionally to its relative size in a population
        
        process sampling
        
        systematic sampling
        
        samples takem from a larger population are selected according to a random starting point and a fixed interval
        
        subgroup sampling
        
        a group of consecutively processed items collected at regular intervals
    - - measure of central tendency
        
        where the values in a set of data are centered
        
        mean
        
        median
        
        mode
      - measured of spread
        
        amonut of spread in values in a set of data
        
        range
        
        difference between smallest and largest
        
        variance
        
        standard deviation
    - - histograms
        
        recommendation is have at least 45 data points so that pattern of variation can be displayed properly
      - box plots
        
        can create with as little as 8-10 data points
        
        can display multiple box plots on same graph
    - - attribute agreement analysis (discrete data)
        
        percent agreement between appraisers (reproducibility)
        
        percent agreement within appraisers (repeatability)
        
        percent agreement with the expert
        
        all agreements should be 90% or higher otherwise measurement system should be improved before used
      - gage repeatability and reproducibility (continuous data)
    - - DPU (defects per unit)
      - DPO (defects per opportunity)
      - DPMO (defects per million opportunities)
      - RTY (rolled throughput yeild)
        
        measures end-to-end process performance
      - COQ (cost of quality)
        
        1) prevention
        
        2) appraisal
        
        3) internal failure
        
        4) external failure
      - COPQ (cost of poor qualitty)
        
        internal failures + external failures + appraisals
  - - - hypothesis
        
        null hypothesis
        
        no difference between groups
        
        alternate hypothesis
        
        difference between groups
      - errors
        
        type 1
        
        rejection of the null hypothesis when in fact it is true
        
        alpha = maximum risk threshold for a type 1 error
        
        determining alpha
        
        risk tolerance
        
        application
        
        industry needs
        
        alpha = 0.05 is commonly used
        
        alpha aka significance level
        
        type 2 error
        
        failing to reject the null hypothesis whne it is in fact false
        
        beta = probability of type 2 error
      - p-value
        
        probability of obtaining a value for the test statistic that is as extreme, or more extreme than what is observed, given that the null hypothesis is true
        
        if p-value <= alpha then reject the null hypothesis
        
        "when the p-value is low then the null must go"
        
        if p-value > alpha then fail to reject the null hypothesis
      - power = 1 - beta
      - confidence level = 1 - alpha
    - - one sample T-Test
      - two sample T-Test
      - anova
    - - one sample variance test
        
        bonett method
        
        chi-squared method
      - equal variance test
        
        bartlett's test
        
        normal data
        
        levene's test
        
        non-normal data
    - - one-proportion test
      - two proportions test
    - - chi-squared test for independance
        
        aka chi-squared test for association
    - - correlation
        
        connection or relationship between two variables
        
        correlation coefficient
        
        number that quantifies the strength of the relationship
        
        ranges from -1 to +1
        
        regression equation
        
        equation of the relationship between selected values of one variable (X) and observed values of the other (Y)
  - - - factors/independent variables
      - dependent variable
      - makes it possible to test multiple factors
    - - definition
        
        systematic methodology for the design or redesign of products, processes, and services
      - characteristics
        
        1) quality is "designed in"
        
        2) customer expectations/needs are prioritized from the start
        
        3) requirements flow down (link between customer requirements and process requirements)
        
        4) quality is predictable
      - methodology followed is DMADV
        
        define
        
        establish business case for desig nprojects
        
        approve project charter
        
        measure
        
        analyze the voice of the customer
        
        identify and prioritize critical-to-quality (CTQ) requirements
        
        analyze
        
        examine CTQs to determine functionality needed
        
        design
        
        create product features
        
        finalize specifications
        
        develop process requirements
        
        optimize parameters
        
        verify
        
        final design is verified
        
        testd for producibility
        
        meets customer needs
        
        design controls
        
        implement
  - - - focuses on varation, shown over time
        
        random variation
        
        due to change, common causes
        
        e.g.
        
        poor design
        
        bad methods
        
        lack of training or supervision
        
        if only random variation exists then process is said to be "in control"
        
        nonrandom variation
        
        due to sporadic, assignable, or special causes
        
        e.g.
        
        jammed equipment
        
        broken tools
        
        contaminations
        
        untrained temp workers
        
        if nonrandome variation exists then process is said to be "out of control"
      - plot e.g.
        
        product dimensions
        
        processing times
        
        number of occurrneces
        
        number of defets
        
        proportion of defects
      - SPC charts for variables / continuous data
        
        chart types
        
        Individual/Moving Range (I-MR chart)
        
        Individual Charts (I-Chart)
        
        Xbar-R charts
        
        Xbar-S charts
        
        out of control conditions
        
        a point outside the control limits
        
        9 points in a row on same side of the centerline
        
        6 points in a row, all increasing or decreasing
        
        14 points in arow, alternating up and down
        
        2 out of 3 points more tha n2 standard deviations from the centerline
        
        4 out of 5 points more than 1 standard deviation from the centerline
        
        15 successive points within 1 standard devisation of centerline
        
        8 success points more than 1 standard deviation from cetnerline
      - SPC charts for attributes / discrete data
        
        chart types
        
        P-chart
        
        plots the proportion of defective units
        
        NP-chart
        
        plots the number of defectuve units
        
        C-chart
        
        plots the number defects
        
        U-chart
        
        plots the number of defect per unit
        
        out of control conditions
        
        any point outside the control limits
        
        9 points in arow on same side of the centerline
        
        6 points in arow, all increasing or decreasing
        
        14 point in a row, alternating up and down
- - - - your job is to translate VOCs to CTQs
        
        don't want to wait to long (VOC)
        
        on time delivery (CTQ)
        
        order-to-delivery time in minutes (measurement)
        
        don't want cold pizza (VOC)
        
        hot pizza when delivered (CTQ)
        
        temperature of pizza in degrees F (measurement)
      - needs and expectations expressed in customer's language
    - - performance characteristics of a process, product or service that are critically important to customers
      - are measureable
        
        how good need to be in order to satisfy customers
  - - - per unit, how many things could possibly go wrong
      - increases as steps/complexity of creating that unit increases
    - - total number of defects / total number of defect opportunities
  - - - is the number of standard deviations between the mean and the nearest specification limit
    - - 6 = no more than 3.4 DPMO (which equals 99.99966% good)
      - 5 = no more than 233 DPMO (which equals 99.9767% good)
      - 4 = no more than 6210 DPMO (which equals 99.379% good)
      - 3 = no more than 66807 DPMO (which equals 93.3193% good)
      - 2 = no more than 308537 DPMO (which equals 69.1463% good)
      - 1 = no more than 690000 DPMO (which equals 31.00% good)
- - - - effectiveness
      - efficiency
  - - - strive
        
        perfection in achieving
        
        effectiveness
        
        efficiency
        
        customer and business requirements
        
        proactive
        
        prevention-based
    - - level of quality near perfection
      - defect level
        
        no more than 3.4 parts per million (no more than six sigma standard deviation)
        
        process is very consistent, with very little variation, and therefore has a very small standard deviation
    - - DMAIC
- - - - approves project charter
        
        defines the project
        
        timeline
        
        objective
        
        scope
      - selects project team
      - launches the project
      - provides time and resources to work on project
        
        reassigns and distributes work
      - monitors progress
      - removes roadblocks
      - approves improvements
        
        process changes
        
        key metrics
        
        procedures
        
        training
        
        job descriptions
      - signs off on project completion
- - - - examine company annual goals
        
        how will they be achieved
        
        how can six sigma be applied
        
        what improvements are needed
      - identify performance gaps in key metrics
        
        do these performance gaps have to be closed
      - review performance reports
        
        look for shortfalls
      - analyze customer complaint data
        
        consider
        
        warranties
        
        returns
        
        credits
        
        these indicate deficiencies
      - conduct an audit
        
        customer
        
        internal
        
        any recurring findings?
      - solicit employee suggestions
        
        especially if supported by data
    - - recurring or chronic issue
      - specific, to a particular
        
        process
        
        product
        
        service
      - measureable
        
        cannot improve what cannot measure
      - have a significant impact
        
        operational
        
        and financial
      - in alignment with goals
        
        company annual goals
        
        established by senior management
    - - should be done by senior management
        
        because they can see entire forest, not just trees
      - put company's limited bandwidth to the best possible use
      - consider
        
        financial impact
        
        degree of urgency
        
        customer impacts
        
        difficulty of execution
        
        level of change required by the organization
  - - - organizes and supports the project
      - typically a senior manager
        
        has an interest or stake in the project
        
        select manager who's goal will be one step closer if project succeeds
        
        so that has enough
        
        clout to authorize time and resources needed for the project
        
        influence to remove any roadblocks
      - selected by senior management team of that facility, division, business unit
    - - leads the team
      - executes the project
      - has knowledge and experience in the process
      - can lead a cross-functional team
      - selected by project champion
    - - select the best people for the project
      - selected by champion
        
        with help of project leader
- - - - more wip there is, more waiting there is
      - items being processed andthose waiting to be procesed
    - - rate at which customer leaves
    - - time between the start and completion of a process
      - how used depends on context
        
        could look at one machine
        
        could look at whole production system
    - - takt is german for musical beat e.g. of a drum
      - takt time
        
        time between units that customers demand
  - - - n = (Z subscript(0.95) s / d)squared
      - sample size for mean
    - - sample size for proportion
      - n = (1.96 / d)squared (p) (1-p)
  - - - observed process variation
        
        part-to-part variation
        
        measurement system variation
        
        variation due to gages
        
        repeatability
        
        bias
        
        stability
        
        linearity
        
        variation due to operators
        
        reproducability
        
        operator
        
        operator-part interation
        
        location variations
        
        bias
        
        difference between observed mean and the actual value
        
        stability
        
        measurements that do not change over time
        
        linearity
        
        difference in bias in a gage's operating range
        
        spread variations
        
        repeatability
        
        ability to consistently repeat the same results from multiple mesurements of the saem item by the same person using the same tools undeer the same conditions
        
        reproducibility
        
        ability to consistenyly reproduce the same result across multiple individuals measureing the same item using the same tools under the same conditions
  - - - CP = specification width / process spread
    - - CPK = munimum of (USL - Mean)/3s and (Mean - LSL)/3s
    - - (USL - Process mean / Half of process spread)
    - - (Process mean - LSL / Half of process spread)