Probability Distributions

Continuous Improvement Toolkit . www.citoolkit.com
Continuous Improvement Toolkit
Probability Distributions

The Continuous Improvement Map
Check Sheets
Data
Collection
Process MappingFlowcharting
Flow Process Charts**
Just in Time
Control Charts
Mistake Proofing
Relations Mapping
Understanding
Performance**
Fishbone Diagram
Design of Experiment
Implementing
Solutions***
Group Creativity
Brainstorming Attribute Analysis
Selecting & Decision Making
Decision Tree
Cost Benefit Analysis
Voting
Planning & Project Management*
Kaizen Events
Quick Changeover
Managing
Risk
FMEA
PDPC
RAID Log*
Observations
Focus Groups
Understanding
Cause & Effect
Pareto Analysis
IDEF0
5 Whys
Kano
KPIs
Lean Measures
Importance-Urgency Mapping
Waste Analysis**
Fault Tree Analysis
Morphological Analysis
Benchmarking***
SCAMPER***
Matrix Diagram
Confidence Intervals
Pugh Matrix
SIPOC*
Prioritization Matrix
Stakeholder Analysis
Critical-to Tree
Paired Comparison
Improvement Roadmaps
Interviews
Quality Function Deployment
Graphical Analysis
Lateral Thinking
Hypothesis Testing
Visual Management
Reliability Analysis
Cross Training
Tree Diagram*
ANOVA
Gap Analysis*
Traffic Light Assessment
TPN Analysis
Decision Balance Sheet
Risk Analysis*
Automation
Simulation
Service Blueprints
DMAIC
Product Family MatrixRun Charts
TPM
Control Planning
Chi-Square
SWOT Analysis
Capability Indices
Policy Deployment
Data collection planner*
Affinity DiagramQuestionnaires
Bottleneck Analysis
MSA
Descriptive Statistics
Cost of Quality*
Process Yield
Histograms 5S
Pick Chart
Portfolio Matrix
Four Field Matrix
Root Cause Analysis Data Mining
How-How Diagram***Sampling
Spaghetti **
Mind Mapping*
Project Charter
PDCA
Designing & Analyzing Processes
CorrelationScatter Plots Regression
Gantt Charts
Activity NetworksRACI Matrix
PERT/CPMDaily Planning
MOST
Standard work Document controlA3 Thinking
Multi vari Studies
OEE
Earned Value
Delphi Method
Time Value Map**
Value Stream Mapping**
Force Field Analysis
Payoff Matrix
Suggestion systems Five Ws
Process Redesign
Break-even Analysis
Value Analysis**
FlowPull
Ergonomics

 Most improvement projects and scientific research studies are
conducted with sample data rather than with data from an
entire population.
- Probability Distributions

What is a Probability Distribution?
 It is a way to shape the sample data to make predictions and
draw conclusions about an entire population.
 It refers to the frequency at which some events or experiments
occur.
Can height

 Helps finding all the possible values a random variable can take
between the minimum and maximum possible values.
 Used to model real-life events for which the outcome is
uncertain.
 Once we find the appropriate model,
we can use it to make inferences
and predictions.
POPULATION
Sample

 Line managers may use probability distributions to generate
sample plans and predict process yields.
 Fund managers may use them to determine the possible
returns a stock may earn in the future.
 Restaurant mangers may use them to resolve
future customer complaints.
 Insurance managers may use them to forecast
the uncertain future claims.

 Probability runs on a scale of 0 to 1.
 If something could never happen, then it has a probability of 0.
• For example, it is impossible you could breathe and be under
water at the same time without using a tube or mask.

 If something is certain to happen, then it has a probability of 1.
• For example, it is certain that the sun will rise tomorrow.

 You might be certain if you examine the whole population.
 But often times, you only have samples to work with.
 To draw conclusions from sample data, you should compare
values obtained from the sample with the theoretical values
obtained from the probability distribution.

 There will always be a risk of drawing false conclusions or
making false predictions.
 We need to be sufficiently confident before taking any decision
by setting confidence levels.
• Often set at 90 percent, 95 percent or 99 percent.

 Many probability distribution can be defined by factors such as
the mean and standard deviation of the data.
 Each probability distribution has a formula.

 There are different shapes, models and
classifications of probability distributions.
 They are often classified into two
categories:
• Discrete.
• Continuous.

Discrete Probability Distribution:
 A Discrete Probability Distribution relates to discrete data.
 It is often used to model uncertain events where the possible
values for the variable are either attribute or countable.
 The two common discrete probability distributions are Binomial
and Poisson distributions.

Binary Distribution:
 A discrete probability distribution that takes only
two possible values.
 There is a probability that one value will occur
and the other value will occur the rest of the time.
 Many real-life events can only have two possible
outcomes:
• A product can either pass or fail in an inspection test.
• A student can either pass or fail in an exam.
• A tossed coin can either have a head or a tail.
 Also referred to as a Bernoulli distribution.
0 1
60%
40%

Binomial Distribution:
 A discrete probability distribution that is used for data which
can only take one of two values, i.e.:
• Pass or fail.
• Yes or no.
• Good or defective.
 It allows to compute the probability of the
number of successes for a given number of trials.
• Each is either a success or a failure, given the
probability of success on each trial.
 Success could mean anything you want to consider
as a positive or negative outcome.
Defective
Pass
Pass
Pass
Fail
Fail
Pass
Pass
Pass
Fail

 Assume that you are tossing a coin 10 times.
 You will get a number of heads between 0 and 10.
 You may then carry out another 10 trials, in which
you will also have a number of heads between
0 and 10.
 By doing this many times, you will have a data set which has the
shape of the binomial distribution.
 Getting a head would be a success (or a hit).
 The number of tosses would be the trials.
 The probability of success is 50 percent.

 The binomial test requires that each trial is independent from
any other trial.
 In other words, the probability of the second trial is not affected
by the first trial.

 This test has a wide range of applications, such as:
• Taking 10 samples from a large batch which is 3 percent defective
(as past history shows).
• Asking customers if they will shop again in the next 12 months.
• Counting the number of individuals who own more than one car.
• Counting the number of correct answers in a multi-choice exam.

 The binomial distribution is appropriate when the following
conditions apply:
• There are only two possible outcomes to each trial (success and
failure).
• The number of trials is fixed.
• The probability of success is identical for all trials.
• The trials are independent (i.e. carrying out one
trial has no effect on any other trials).

 The probability of ‘r’ successes P(r) is given by the binomial formula:
P(r) = n!/(r!(n-r))! * pr(1-p)n-r
p: probability of success
n: number of independent trials
r: number of successes in the n trials
The binomial distribution is fully defined if we know both ‘n’ & ‘p’

 The data can be plotted on a graph.
The exact shape of a particular distribution depends on
the values of ‘n’ and ‘p’
0.4
0.3
0.2
0.1
0.0
0 1 2 3 4 5
P(r)

Binomial Distribution – Example:
 In a sample of 20 drawn from a batch which is 5% defective,
what is the probability of getting exactly 3 defective items?
P(r) = n!/(r!(n-r))! * pr(1-p)n-r
P(r) = 20!/(3!(20 - 3)!) * 0.053(1 - 0.05)20-3
P(r) = 1,140 * 0.053 * (0.95)17 = 0.060

Binomial Distribution – Example:
 When sampling, we commonly want to accept a batch if there
are (say) 1 or less defective in the sample, and reject it if there
are 2 or more.
 In order to determine the probability of acceptance, the
individual probabilities for 0 and 1 defectives are summed:
• P (1 or less) = P(0) + P(1)
= 0.358 + 0.377 = 0.735
 So the probability of rejection is:
• 1 – 0.735 = 0.265

Binomial Distribution – Other Examples:
 What is the probability of obtaining exactly 2
heads in the 5 tosses?
 What is the probability that in a random sample
of 10 cans there are exactly 3 defective
units, knowing that on average there is a 5%
defective product?
 What is the probability that a random sample of
4 units will have exactly 1 unit is defective,
knowing that that process will produce 2%
defective units on average?

 In Excel, you may calculate the binomial probabilities using the
BINOM.DIST function. Simply write:
=BINOM.DIST(number of successes, number of
trials, probability of success, FALSE)

 Suppose that we want to know the chance of getting exactly 4
heads out of 10 tosses.
 Instead of using the binomial formula, we might skip straight to
the Excel formula.
 In Excel, we simply write:
 The result value will be 0.205078125.
 This means that there is a 20.5% chance that
10 coin tosses will produce exactly 4 heads.
=BINOM.DIST(4,10,0.5,FALSE)

Hypergeometric Distribution:
 A very similar to the binomial distribution.
 The only difference is that a hypergeometric distribution does
not use replacement between trials.
 It is often used for samples from relatively small populations.

Hypergeometric Distribution:
 Assume that there are 5 gold coins and 25 silver coins in a box.
 You may close your eyes and draw 2 coins without replacement.
shape of the hypergeometric distribution.
 You can then answer questions such as:
• What is the probability that you will draw one
gold coin?
Note that the probability of success on each trial
is not the same as the size of the remaining
population will change as you remove the coins.

Poisson Distribution:
 It is not always appropriate to classify the outcome of a test
simply as pass or fail.
 Sometimes, we have to count the number of defects
where there may be several defects in a single item.
 The Poisson Distribution is a discrete probability
distribution that specifies the probability of a certain
number of occurrences over a specified interval.
• Such as time or any other type of measurements.
Defects
14
8
11
2
13
17
11
9
12

 With the Poisson distribution, you may examine a unit of
product, or collect data over a specified interval of time.
 You will have a number of successes (zero or more).
 You may then repeat the exercise, in which you will also have a
number of successes.
shape of the Poisson distribution.

 This test has a wide range of applications, such as:
• Counting the number of defects found in a single product.
• Counting the number of accidents per year in a factory.
• Counting the number of failures per
month for a specific equipment.
• Counting the number of incoming calls
per day to an emergency call center.
• Counting the number of customers who
will walk into a store during the holidays.
X X
X

 The Poisson distribution is appropriate when the following
conditions apply:
• Occurrences take place within a defined interval or area of
opportunity (per sample, per unit, per hour, etc.).
• Occurrences take place at a constant rate, where the rate is
proportional to the length or size of the interval.
• The likelihood of occurrences is not affected by
which part of the interval is selected (e.g. the
time of day).
• Occurrences take place randomly, singly and
independently of each other.

 The probability of ‘r’ occurrences is given by the Poisson formula:
P(r) = λr e-λ / r!
λ: average or expected number of
occurrences in a specific interval
r: number of occurrences
The Poisson distribution is fully defined if we know the value of λ

 The data can be plotted on a graph.
The exact shape of a particular distribution depends solely on the
value of λ
r P(r)
0 0.607
1 0.303
2 0.076
3 0.013
4 0.002
5 <0.001
6 0
0.6
0.5
0.4
0.3
0.2
0.1
0 1 2 3 4 5
P(r)

Poisson Distribution – Example:
 What is the probability of at least 1 accident taking place in a
given week?
 In other words, what is the probability of
1 or more accidents taking place?
 We can add the probabilities above 0
or use the complement method.
 P(1 or more accidents) = 1 – P (0 accidents)
 = 1 – 0.607 = 0.393
r P(r)
0 0.607
1 0.303
2 0.076
3 0.013
4 0.002
5 <0.001
6 0

Poisson Distribution – Other Example:
 What is the probability of assembling 1 part with less than 3
defects, knowing that a study has determined that on average
there are 3 defects per assembly.

 In Excel, you may calculate the Poisson probabilities using the
POISSON.DIST function. Simply write:
=POISSON.DIST(number of successes, expected
mean, FALSE)

Continuous Distributions:
 Relates to continuous data.
 Can take any value and can be measured with any degree of
accuracy.
 The commonest and the most useful continuous distribution is
the normal distribution.
 There are other continuous probability
distributions that are used to model
non-normal data.

Uniform Distribution:
 All the events have the exact same probability of happening
anywhere within a fixed interval.
 When displayed as a graph, each bar has approximately the
same height.
 Often described as the rectangle distribution.
 Does not occur often in nature.
 However, it is important as a reference distribution.

Exponential Distribution:
 Often used in quality control and reliability analysis.
 Mainly concerned with the amount of time until some specific
event occurs.
• Calculating the probability that a computer part will last less than
six year.
 Its shape is highly skewed to the right.
• There is a much greater area below the
mean than above it.

Normal Distribution:
 A symmetrical probability distribution.
 Most results are located in the middle and
few are spread on both sides.
 Has the shape of a bell.
 Can entirely be described by its mean and
standard deviation.
 Normality is an important assumption when conducting
statistical analysis so that they can be applied in the right
manner.

Bimodal Distribution:
 It has two modes (or peaks).
 Two values occur more frequently than the other values.
 It can be seen in traffic analysis where traffic peaks during the
morning rush hour and then again in the evening rush hour.
XXX
XXX
XXX
XX
XX
X
X
X
X
XX
X
X
XXX
XX
XXX
X
XX
X
X
X
XXX
X
X
X

 It may also result if the observations are taken from two
different populations.
• This can be seen when taking samples from two different shifts or
receiving materials from two different suppliers.
• There is actually one mode for the two data sets.
POPULATION 1 POPULATION 2
Sample

 The bimodal is considered a Multimodal Distribution as it has
more than one peak.
 This may indicate that your sample has several patterns of
response, or has been taken from more than one population.
 Multimodal distributions can be seen in daily water distribution
as water demand peaks during different periods of the day.

 There are different shapes, models and
classifications of probability distributions.
 It is always a good practice to know the
distribution of your data before proceeding
with your analysis.
 Once you find the appropriate model, you
can then perform your statistical analysis in the right manner.
 Minitab can be used to find the appropriate probability
distribution of your data.
- Probability Distributions in Minitab

 You may use the Individual Distribution Identification in
confirm that a particular distribution best fits your current data.
 It allows to easily compare how well your data fit various
different distributions.
 Let’s look at an example where a hospital is seeking to detect
the presence of high glucose levels in
patients at admission.

 Remember to copy the data from the Excel worksheet and
paste it into the Minitab worksheet.
 To find out the probability distribution that best fit the data:
• Select Stat > Quality Tools > Individual Distribution Identification.
• Specify the column of data to analyze.
• Specify the distribution to check the
data against.
• Then click OK.

 The resulted graph if only the normal distribution has been selected:
Minitab can also create a series of graphs that provide
most of the same information along with probability plots

 A given distribution is a good fit if the data points approximately
follow a straight line and the p-value is greater than 0.05.
A good place to start is to look at the
highest p-values in the session window
In our case, the data does not appear
to follow a straight line

 You may transform your non-normal data using the Box-Cox or
Johnson transformation methods so that it follows a normal
distribution.
 You can then use the transformed data with any analysis that
assumes the data follow a normal distribution.

 You can also use the Probability Distribution Plots to clearly
communicate probability distribution information in a way that can be
easily understood by non-experts.
 Select Graph > Probability Distribution Plot,
and then choose one of the following options:
• View Single to display a single probability distribution plot.
• Vary Parameters to see how changing parameters will affect the distribution.
• Two Distributions to compare the shape of
curves based on different parameters.
• View Probability to see where target values
fall in a distribution.

 Here is an example of a process with a mean of 100, a standard
deviation of 10 and an upper specification limit of 120.
 The following screenshot shows the shaded area under the curve that
is above the upper specification limit:
0.04
0.03
0.02
0.01
0.00
X
120
0.02275
100
Distribution Plot
Normal, Mean=100, StDev=10

Further Information:
 A truncated distribution is a probability distribution that has a
single tail.
 Can be truncated on the right or left.
 Occurs when there is no ability to know about or record data
below a threshold or outside a certain range.

 An edge peak distribution is where there is an out of place peak
at the edge of the distribution.
 This usually means that the data has been collected or plotted
incorrectly, unless you know for sure your data set has an
expected set of outliers.

 A comb distribution is so-called because the distribution looks
like a comb, with alternating high and low peaks.
 A comb shape can be caused by rounding off.
 Or if more than one person is recording the data or more than
one instrument is used.

 Student’s T-Distribution:
• A bell shaped probability distribution that is symmetrically shaped.
• Generally much flatter and wider than the normal distribution (has
more probability in the tails).
• There is a t-distribution for each sample size of n > 1.
• Appropriate when constructing confidence
intervals based on small samples (n<30)
from populations with unknown variances.

 Student’s T-Distribution:
• Defined by a single parameter,
the degrees of freedom (df).
• The degrees of freedom refers
to the number of independent
observations in a set of data.
• It is equal to the number of sample observations minus one.
• The distribution of the t-statistic from samples of size 8 would be
described by a t-distribution having 7 degrees of freedom.
df 0.1 0.05 0.02 0.01 …
1 6.314 12.706 31.821 63.657 …
15 1.753 2.131 2.602 2.947 …
30 1.697 2.042 2.457 2.750 …
∞ 1.645 1.960 2.327 2.576 …
Critical values of the t-distribution

 F-Distribution:
• A continuous probability distribution that will help in the testing of
whether two normal samples have the same variance.
• Used to compute probability values in the analysis of variance
(ANOVA) and the validity of a multiple regression equation.
• An example of a positively skewed
distribution is household incomes.

 F-Distribution:
• Has two important properties:
• It’s defined only for positive values.
• It’s not symmetrical about its mean. instead, it’s positively
skewed.
• The peak of the distribution happens just to the right of zero.
• The higher the F-value after that point,
the lower the curve.

 F-Distribution:
• Like the student’s t-distribution, the probability is determined by
the degrees of freedom.
• Unlike the Student’s t-distribution, the F-distribution is
characterized by two different types of degrees of freedom:
• Numerator (dfn).
• Denominator (dfd).
• The shape depends on the values of the numerator and
denominator degrees of freedom.

Probability Distributions

More Related Content

What's hot (20)

Similar to Probability Distributions (20)

More from CIToolkit (20)

Recently uploaded (20)

Probability Distributions