Analysis Part 2
Analysis Part 2
40
20
0
Pace of Course
Last Class
• Wrapped-up measurement
• Process Capability Measurement
• Project analysis
• Basic Probability
• Probability distribu:ons
3
Today’s Class
• Project analysis continued
• Probability Distributions
• Statistical Inference
• Sampling distributions
• Confidence intervals
• Hypothesis testing
• Correlation
• Midterm Exam
4
Overview and Principles of Six Sigma
1. Understand
4. Identify what to 7. Identify performance 9. Generate possible 11. Institutionalize &
customer & business
measure gaps solutions monitor solution(s)
requirements
2. Complete high-level, 10. Prioritize & select 12. Replicate & share
5. Plan and collect data 8. Validate root causes
as-is process map solution(s) best practices
3. Complete project 6. Determine baseline 13. Celebrate &
charter performance recognize success
Additional Topics
Drawn from the “The DMAIC Roadmap” Advanced Innovation Group Pro Excellence 2018
Overview and Principles of Six Sigma
Step 7 • Brainstorming
• What are the potential causes of the
Identify performance • 5 Why
problem?
gaps • Fishbone diagrams
• Confidence intervals
• t-tests
Step 8 • How do you know these are the root causes? • Correlation & regression
Validate root causes • What are the validated root causes? • ANOVA
• Multi-vari studies
• Design of experiments
Tollgate success: Team understands the root causes and impact of the problem and ready to Improve.
Drawn from the “The DMAIC Roadmap” Advanced Innovation Group Pro Excellence 2018
Overview and Principles of Six Sigma
Step 7 • Brainstorming
• What are the potential causes of the
Identify performance • 5 Why
problem?
gaps • Fishbone diagrams
• Confidence intervals
• t-tests
Step 8 • How do you know these are the
• Correlation & regression
Validate root root causes?
• ANOVA
causes • What are the validated root causes?
• Multi-vari studies
• Design of experiments
Tollgate success: Team understands the root causes and impact of the problem and ready to Improve.
Today!
Drawn from the “The DMAIC Roadmap” Advanced Innovation Group Pro Excellence 2018
Analyze
Probability Distributions
Random Variables
Random Variable
• Numerical description of the outcome of an
experiment
• RVs can be discrete or continuous
9
Random Variables
Random Variable - Example 1
• Experiment: sampling 10 parts and counting
number of defective parts
• Random Variable X: number of defective parts in
sample
Random Variable - Example 2
• Experiment: testing product for pass or fail
• Random Variable Y: 1 if outcome pass, 0 if
outcome fail
10
Probability Distributions
11
Probability Distributions
Cumulative Distribution Function (CDF)
• F(x), specifies the probability that the random
variable X will assume a value less than or
equal to a specified value, x
• P(X ≤ x)
• The probability that the random variable X is
less than or equal to x
12
Important Distributions
• Discrete
• Binomial
• Poisson
• Continuous
• Normal
• Exponential
13
Discrete Distributions
Binomial Distribution
• Describes probability of obtaining exactly x
“successes” in a sequence of n identical
experiments, called trials
• A success:
• Can be any one of two possible outcomes
(binomial) of each experiment
• In some situations, it might represent a
defective item, in others, a good item
• Probability of success in each trial, p
14
Discrete Distributions
Binomial Distribution
• p - probability of a success
• n - number of trials (items) in the sample
• x - number of successes for which the
probability is desired (0, 1, 2, ..., n)
15
Discrete Distributions
Binomial Distribution
• p - probability of a success
• n - number of trials (items) in the sample
• x - number of successes for which the
probability is desired (0, 1, 2, ..., n)
16
Binomial Distribution
• p - probability of a success
• n - number of trials (items) in the sample
• x - number of successes for which the
probability is desired (0, 1, 2, ..., n)
Binomial Distribution
Binomial Distribution
Example 5.4
19
Discrete Distributions
Poisson Distribution
• Related to binomial distribution
• When n is large relative to p, the Poisson
distribution can be used as an approximation
of the binomial
• 𝛌 = expected value or average number of
occurrences
• x = 0, 1, 2, 3, …
• e = 2.71828…
20
Poisson Dist. Excel
• POISSON.DIST(x, mean, cumulative)
21
Continuous Distributions
22
Continuous Distributions
• For continuous RVs, doesn’t make sense to
define a probability for a specific value of x
because there are an infinite number of them
• Probabilities are only defined over intervals
• The cumulative distribution function, F(x),
represents the probability P(X ≤ x)
23
Continuous Distributions
Normal Distribution
• “Bell Curve”
• Symmetric
• Data often assumed normally distributed
• Averages are normally distributed
25
Continuous Distributions
“Standard” Normal Distribution
• μ=0
• σ2 = 1
• z used to represent this particular RV
26
Standardizing Normal RVs
• If x is any value from a normal distribution
with mean μ and standard deviation σ, we may
easily convert it to an equivalent value from a
standard normal distribution using the following
formula:
27
Continuous Distributions
28
Normal Dist. in Excel
• Excel function NORM.DIST(x, mean, standard
deviation, true)
• Calculates cumulative probability F(x) for a
specified mean and standard deviation
• The Excel function NORM.S.DIST(z)
• Calculates cumulative probability for standard
normal distribution as a function of z
29
Normal Distribution
Example 5.6
• Mean number of days (X) between MRI
equipment malfunctions is 1,020
• Standard deviation of 20 days
• Assuming a normal distribution, what is
probability that number of days between
adjustments will be:
• Less than 1,044 days?
• More than 980 days?
• Between 980 and 1044 days?
30
Normal Distribution
Example 5.6
• Less than 1,044 days?
• P(X ≤ 1044)? Use Appendix A
31
Normal Distribution
Example 5.6
• More than 980 days?
32
Normal Distribution
Example 5.6
• More than 980 days?
• This can be found using Excel:
• 1 – NORM. DIST(980, 1020, 20, TRUE)
33
Normal Distribution
Example 5.6
• Between 980 and 1044 days?
34
Normal CDF
Normal Inverse in Excel
• NORM.INV(probability, mean, standard_dev)
• Returns value of x for a given % of CDF
• Probability is cumulative probability
35
Exponential Distribution
36
Exponential Distribution
• Exponential PDF
• Exponential CDF
37
Exponential Dist. in Excel
• The Excel function EXPON.DIST(x, lambda, true)
can be used to compute cumulative
exponential probabilities.
38
Analysis
Statistical Inference – Sampling Distributions
Statistical Inference
• Allows us to make probability statements about
the relationship between sample statistics and
population parameters and draw inferences
• To do this, we first need to understand sampling
distributions
40
Sampling Distributions
• Sample statistics such as 𝑥,̅ s, and p are random
variables that each have their own probability
distribution, mean, and variance
• Thus, different samples produce different
estimates of the population parameters
• These probability distributions are called
sampling distributions.
• Sampling distribution:
• The distribution of a statistic for all possible
samples of a fixed size
• In quality applications, 𝑥̅ and p of most interest
41
Sampling Distn. of Mean
• Sampling distribution of mean
• Expected value of the sample mean is the
population mean
• Standard deviation of the sample mean (called
the standard error of the mean) is the
population standard deviation divided by the
square root of the sample size
42
Central Limit Theorem
• Extraordinary theorem that makes the normal
distribution so powerful
43
CLT
Illustration
44
Central Limit Theorem
Example 5.9
• The mean length of shafts produced on a lathe
has historically been 50 inches, with a standard
deviation of 0.12 inches
• If a sample of 36 shafts is taken, what is the
probability that the sample mean is greater than
50.04 inches?
45
Sampling Distn. of Mean
Example 5.9
• If a sample of 36 shafts is taken, what is the
probability that the sample mean is greater than
50.04 inches?
• μ = 50, standard deviation of 0.12
47
Sampling Dist. of p
Example 5.10
• On average, 15 percent of the batches produced
do not meet specifications
• If 35 batches are tested, what is the probability
that 9 or fewer batches will be rejected?
• Sample proportion is p = 9/35 = 0.25714
• Sampling distribution of p has a mean 𝜋 of =
0.15 and standard deviation of:
51
Common CIs
55
CI Examples
Example 5.11
• Temperature of sterilizer has a mean that
exceeds 100°C and meets their requirements
56
CI Examples
Example 5.12
• Same situation as Ex. 5.11, but population
standard deviation is unknown
• Sample of n = 16 was taken
• Sample mean was found to be 𝑥̅ = 100.3
• Sample standard deviation s = 0.7
• Find 95% CI for population mean
57
CI Examples
Example 5.12
• Find 95% CI for population mean
58
CI Examples
Example 5.12
• Target temperature not in the 95% CI
• Why different result?
59
CI Examples
Example 5.12
• Target temperature not in the 95% CI
• Why different result?
• t-distribution has fatter tails
60
Analysis
Statistical Inference – Hypothesis Testing
Hypothesis Testing
• Involves drawing inferences about two
contrasting propositions (hypotheses) about
value of a population parameter
• One proposition is hypothesized to be true in
the absence of contradictory data
• The null hypothesis, H0
• Other accepted if null hypothesis rejected
• Alternative hypothesis, H1
62
Hypothesis Testing
• Involves the following steps
1. Formulate a hypothesis to test
2. Select level of significance of test
3. Determine decision rule on which to base a
conclusion
4. Collect data & calculate a test statistic
5. Apply decision rule to test statistic & draw
conclusion
63
Level of Significance (𝛼)
• 𝛼 – alpha
• Risk we are willing to take in falsely rejecting
the null hypothesis, when it is in fact true
• AKA Type I Error: the probability of rejecting
the null hypothesis when the null hypothesis
is true
• AKA Significance level of test
• Commonly used levels for α are 0.10, 0.05, and
0.01
64
Level of Significance (𝛼)
• Drawing in class
65
Decision Rule
66
Critical Value
• Determined by:
• null hypothesis
• chosen level of significance
• Divides a sampling distribution between
rejection region and non-rejection region
• One-tailed test: rejection region is either in
upper or lower tail with probability α
• Two-tailed test: rejection region is in both tails
with area of α/2
67
Critical Value
• Drawing in class
68
Exam Format
§ Multiple Choice
ü Some theory and conceptual questions
ü Some calculations and interpretations
§ Short Answer
ü Some theory including figures
ü Some calculations
Main Points for Midterm Exam
§ Make Sure to Arrive Early
ü Please arrive at 5:30PM
§ What to Study
ü Some topics are more important than others.
ü Spend your time on the main topics covered in lectures and slides.
ü Don’t waste time on topics we haven’t emphasized in class.
§ How to Prepare for the midterm
ü Make a list of your problem areas.
ü Eliminate topics/problems not mentioned in lecture slides.
ü Use the lecture slides as a guide.
ü Read the relevant textbook chapters on the covered topics.
ü Practice questions at the end of chapters.
ü Bring a calculator on the day of the exam.
ü A formula sheet will be provided.
71