Histograms

Purpose:  Todetermine the spread or variation of a set of data

points in a graphical form.
Variation: It
is always a desire to produce things that are equal
to their design values.  For example if we are producing a
number of cylindrical objects having a certain diameter value,
we wish that every part produced will have the same diameter
value.  But this is not always the case.  We will always have
variation in the values of the diameter between each produced
part. Variation is everywhere.  It is found in the output of any
process: manufacturing, service, or administrative.  But
variation is not all bad.  One characteristic of variation is that
it always displays a pattern, a distribution.  These patterns can
tell us a great deal about the process itself and the causes of
problems found in the process.  Histograms help us identify
and interpret these patterns.

Figure 1

Histograms: Ahistogram is a tool for summarizing, analyzing,

and displaying data.  It provides the user with a graphical
representation of the amount of variation found in a set of
data.  Histograms sort observations or data points, which are
measurable data, into categories and describes the frequency
of the data found in each category.

Constructing a Histogram:  The following are the steps followed in

the construction of a histogram: The following are the steps
followed in the construction of a histogram: 
Data collection:  To ensure good results, a minimum of 50 data points, or
samples, need to be collected
Calculate the range of the sample data:  The range is the difference between the
largest and smallest data points. Range = Largest point - smallest point.
Data points need to be divided on the X axis into classes, look at Figure 1above. 
This is a very important step because how you select the class scale will
determine the effectiveness of the histogram in interpreting the variation found
in the data set.  Table 1 below lists some of the rules of thumb for determining
the number of classes to use with respect to the number of collected data points.

Table 1. Rules of thumb for class selection.

Sample No# of
size classes
0-50 5-7
51-99 8-10
100-250 10-15
over 250 15-20

 
Calculate the size of the class interval. The class interval is the width of each
class on the X axis.  It is calculated by the following formula:  Class interval =
Range / Number of classes.
Determine the class boundary.  They are the largest and smallest data points that
can be included in each class.
Calculate the number of data points (frequency) that are in each class.  A tally
sheet is usually used to find the frequency of data points in each interval.
Draw the Histogram, as in Figure 1, and plot the data.
An Example: Company X manufactures small resistors with a
resistance value of 100 Ohms.  Recently, the company has
been receiving numerous complaints from their customers
about these resistors.  The value of the resistance has been
deviating from the design value.  A team of workers was
selected to investigate this deviation in the resistance values. 
As a starting point, the team decided to construct a histogram
to see the extent of variation in the resistance value.  A sample
of 50 resistors from the production line was taken for the
construction of the histogram.  The sample's resistance values
are listed in Table 2 below.
Table 2. The resistance value for 50 resistors.
87 90 86 101 83
84 108 89 80 85
100 79 90 92 90
81 94 86 85 100
83 110 90 100 91
91 89 100 81 107
76 74 99 85 85
76 100 74 79 77
90 85 91 84 100
85 82 88 85 99

-The range of the fifty data points = 110 - 74 = 36.

-The Number of classes is 6 ( from Table 1).
-The class interval = range/number of classes= 36/6 = 6 
-The class boundary and the frequency of each class are listed
in Table 3 below.
Figure 2

Table 3.
Cla Bounda Tot
Tally
ss ries al
1 71-77 ///// 5
2 78-83 //////// 8
/////////////
3 84-90 18
/////
4 91-97 ///////// 9
5 98-104 /////// 7
6 105-111 /// 3

-Figure 2 above shows the histogram of the 50 resistor sample

taken from the production line.  It is clear that the majority of
resistors, about half, have a resistance value between 78 and
90 Ohms.  This makes us conclude that company X's original
assumption that the resistors have a resistance value of 100
Ohms is incorrect.  The nest step is to look at the
manufacturing process and determine the cause or causes of
this deviation.
Conclusion:  Histogram are simple tools that allow the user to
identify and interpret the variation found in a set of data
points.  They are used to summarize and display data in a
simple but clear manner.  It is important to remember that
histograms do not give solutions to problems.  They only
provide a starting point for the improvement process.  Also,
the results obtained from any histogram will depend on the
date which the histogram is made.  If the data is inaccurate
then any result obtained from the histogram is also inaccurate.

Flow Charts
Purpose:  Flow Charts provide a visual illustration of the sequence of operations
required to complete a task.

Flow Charts:  Flow
charting
is the first step we take in
understanding a process. 
Whether this process is
an administrative or a
manufacturing one, flow
charts provide a visual
illustration, a picture of
the steps the process
undergoes to complete it's
task.  From this picture
we can see how this process and the elements comprising it,
fit into the overall picture of the business.  Every process will
require input(s) to complete it's task, and will provide
output(s) when the task iscompleted.  For example, an
injection molding machine requires inputs in the form of raw
material and proper machine settings to be able to produce a
proper part.  The output of the injection molding process is the
finished part. (See above picture)  Flow charts can be drawn
in many styles.  They can be drawn by using pictures,
engineering symbols, or just squares and rectangles.  Also,
flow charts can be used to describe a single process, parts of a
process, or a set of processes.  There is no right or wrong way
to draw a flow chart.  The true test of a flow chart is how well
those who create and use it can understand it.
 Construction of the Flow Chart (General Guide Lines):

* Involving the right people in making the flow chart.  This

includes people who actually do the work of the process,
supervisors, suppliers to the process, and customers to the
process.
* Flow charts usually require more time to construct than
expected.  Therefore, enough time must be allotted to the
group members to finish their work.
* Asking questions is the key to the flow charting process. 
The more questions everyone asks the better.  Here is a list of
helpful questions to ask when constructing a flow chart:
1. What is the first thing that happens?
2. What is the next thing that happens?
3. Where does the output(s) of this operation go?
4. Where does the input(s) to the process come from?
5. How does the input(s) get to the process?
6. Is their anything else that must be done at this point?

A Special Benefit of Flow Charting:   People involved in

constructing a flow chart begin to better understand the
process.  They begin to identify areas for improving the
process.  They also begin to realize how the process and all
the people involved, including them, fit into the overall
process or business.  Because of their involvement, they
become enthusiastic supporters to quality and process
improvement.  Finally, they all begin to understand the
process in the same terms.
Conclusion:   Flowcharts allow the user to have a visual
illustration of a process.  They are the first step in the process
of understanding manufacturing as well as administrative
processes. They encourage team work and employee
involvement.

Scatter Diagrams
Purpose:  To identify correlations that might exist between a quality characteristic
and a factor that might be driving it.

Scatter Diagrams:  A scatter diagram is a nonmathematical or graphical approach

for identifying relationships between a performance measure and factors that might
be driving it.  This graphical approach is quick, easy to communicate to others, and
generally easy to interpret.  Figure 1 below shows a commonly used structure for a
scatter diagram.  The data needed to construct the scatter diagram must be
collected in pairs (X,Y).  Almost always the performance characteristic, Y, is
plotted on the vertical axis, while the suspected correlated factor, X, is plotted on
the horizontal axis.  The point of intersection between the two axes is the average
of each of the sets of data (i.e. the average of all the X's and the average of all the
Y's).  The collected data is not for only observing the quality characteristic under
investigation but also observing other factors or causes that might have an impact
on the quality characteristic.  For example if we are taking measurements of the
surface finish of a machined part, we will want to take measurements of other
factors such as feed rate and tool condition that could have an effect on our surface
finish quality characteristic.

Constructing the Scatter Diagram:

Step 1:  Select the two items you wish to study.  The results of the cause-and-
effect diagram could be very helpful in determining which items to select.  For
example the two items could be an effect and a related cause.
Step 2:  Collect the data.  The more data you have more accurate your
conclusions will be.  Always remember that the type of data needed to construct
the scatter diagram is paired.
Step 3:  Draw the axis of the scatter diagram.  Remember that the performance
characteristic in on the Y axis, and the suspected correlated factor on the X axis. 
The point of intersection of the two axes is the average of each of the sets of
data.  You can also make the origin point (0,0) your intersection point.
 Step 4:  Plot each set of paired data onto the graph (i.e. (Xo,Yo), (X1,Y1),
(X2,Y2),......,(Xn,Yn), where n is the number of samples taken.

Interpreting the Results:  Once all the data points have been
plotted onto the scatter diagram, you are ready to determine
whether their exists a relation between the two selected items
or not. When a strong relationship is present, the change in
one item will automatically cause a change in the other. If no
relationship can be detected, the change in one item will not
effect the other item. Their are three basic types of
relationships that can be detected to on a scatter diagram:

Figure 1

1. Positive relationship; As the item on the X axis increases,

the item on the Y axis also increases, and vice versa, look at
Figure 1.
Figure 2

2. Negative relationship; As the item on the X axis increases,

the item on the Y axis decreases, look at Figure 2.

Figure 3

3. No relationship; Changing the values of item X does not

have any effect on the value of item Y, look at Figure 3.
Conclusion: Scatterdiagrams allow the user to graphically
identify correlations that could exist between a quality
characteristic and a factor that might be driving it.  It is a
quality tool that is simple, easy to communicate to others, and
generally easy to interpret.

Pareto Charts Overview

Purpose:

The purpose of the Pareto chart is to prioritize problems - to

decide what problems must be addressed.  No company has
enough resources to tackle every problem, so they must
prioritize.  The purpose of this report is to inform the reader of
the Pareto principles and how it can be applied to the
manufacturing environment.  I will try to answer commonly
asked questions regarding the Pareto process.
Pareto Principle:

The Pareto concept was developed by the Italian economist

Vilfredo Pareto describing the frequency distribution of any
given characteristic of a population.  Also called the 20-80
rule, he determined that a small percentage of any given group
(20%) account for a high amount of a certain characteristic
(80%).  For instance in the BYU Law parking lot, there are
basically 9 different colors of cars: white, blue, red, black, tan,
silver, brown, yellow, and green.  I decided to illustrate
the Pareto principle by the different colors of cars. All colors
are not created equally. Out of the nine different colors, three
(red, blue, and white) make up about 64% of the cars.  The
same principle can be applied to all aspects of manufacturing. 
The Pareto chart is especially helpful in improving
manufacturing processes.
Where do I begin?
The important thing is to begin.  Start somewhere.  Choose a
process that is not producing the yields you would like it to
produce.  Are there an excess amount of parts that you are
having to rework or scrap?  Why?  What are the reasons for
scrapping those parts?  Make a list of the causes of the
problem.  Don't worry about narrowing the list because it will
narrow itself.  Keep track of the number of scrapped or
reworked parts and what the reason is.  In the example above,
I wanted to know what color of car was most popular.  I could
have also counted cars by their brand, or their type (compact,
subcompact, etc.)  It doesn't really matter.  As the data begins
to come in, then you can make more decisions.
What do I do next?

After you have collected a sufficient amount of data, chart

what you have.  In my example, I used the number of cars. 
This is a great method.  However, you may want to keep track
of the cost of each product failure.  For example, if you have a
milling operation, and scratching the surface of the product
causes the most reworks, but it only costs $2 to correct,
whereas milling an uneven surface, while occurring less often,
is actually more expensive to correct.  That is something that
you must decide, whether to go by cost or total numbers.  If
you don't know which to choose, then do both.
There may be times where different products come off the
same production line.  In this instance, it may be important to
keep charts on the different products.  For instance, there may
be 10 scraps out of 10,000 of product A; whereas with product
B, there may be 8 scraps out of 100.  If you were to group the
whole process together, the problem with product B might not
be discovered.  Be aware of grouping products together that
come from the same process.
Now you should have an idea of what problems are most
troublesome.  There should be two or three problems that
stick out.  Focus on those problems causing the greatest
number of reworks or those problems which are the costliest.
After you feel you have improved the process, do
another Pareto chart and find out if your improvements
worked.  If they did, great! and your Pareto chart shows some
new problems that must be worked on.  Continually improve
on the problem that is causing the greatest harm.
Conclusion

You can't choose the wrong process or the wrong problem to

run a Pareto chart on.  The most important thing in improving
quality is to start somewhere, doing something.  As you begin
using the Pareto chart to decide where your problems are, you
will discover many things about your processes and will come
because you will know where to improve.

Cause-and-Effect Diagrams
Purpose:

One important part of process improvement is continuously

striving to obtain more information about the process and it's
output.  Cause-and-effect diagrams allow us to do not just
that, but also can lead us to the root cause, or causes, of
problems.  It is a tool that enables the user to set down
systematically a graphical representation of the trail that leads
ultimately to the root cause of a quality concern or problem.
Cause-and-Effect Diagram:

First developed in 1943 by Mr. Ishikawa at the University of

Tokyo, the cause-and-effect diagram, also known as the
fishbone diagram, relates the symptom or problem under
question to the factors or causes driving it.  It accomplishes
this through a hierarchical relationship between the effect, the
main causes of this effect, and their subsequent relationship to
the sub causes.  While a cause-and-effect diagram can be
developed by an individual, it is best when used by a team.  A
cause-and-effect diagram consists primarily of two sides.  The
right side, effect side, lists the problem or the quality concern
under question.  While the left side, cause side, lists the
primary causes of the problem.  The right hand side can also

include a desired effect the user wishes to achieve.  The

picture below is an example of a eause-and-effect diagram. 
The symptom or Quality Concern is the "Black Spots" that are
occurring in an injection molding process.  The main causes
that could give rise to the problem are the machine, the raw
material, the method or process procedures, and the human
element.  Each of these main causes is composed of sub
causes that influence or effect the quality concern under
question.  The selection of the headings for the main causes
could be generic, such as Man or Method, or they can be
specific such as Training or Wrong Speed.  The important
thing is to continually define and relate causes to each other.
Constructing the Cause-and-Effect Diagram:

Step 1:  Select the team members.  It is preferable that the

team members are knowledgeable about the quality concern
under question.  One of the team members is selected as the
team leader (Facilitator).  His job is to listen to the ideas
presented by the other team members, capture those thoughts
in simple words and write them on the chart.  Also to keep the
team members focused on the problem under investigation.
Step 2:  Assuming that the quality concern has been
identified, write this concern as a problem statement on the
right hand side of the page, and draw a box around it with an
arrow running to it.  This quality concern is now the effect.
Step 3:  Brain-storming.  The team members generate ideas as
to what is causing the effect.  Main causes as well as sub
causes are identified.  As indicated earlier, the team could
choose generic or specific headings for describing the main
causes.
Step 4:  This step could be combined with step 3.  Identify, for
each main cause, its related sub-causes that might effect our
quality concern or problem (our Effect).  Always check to see
if all the factors contributing to the problem have been
identified.  Start by asking why the problem exists.  If a cause
is identified, repeat the question again.  When no other causes
can be identified, you have likely identified all the possible
causes to the problem.
Step 5:  Focus on one or two causes for which an
improvement action(s) can be developed using other quality
tools such as Pareto charts, check sheets, and other gathering
and analysis tools.  This will depend on the team members
themselves on where and how to start.  Agreement between
the team members must be based on consensus.
Conclusion:

Improvement requires knowledge.  The more information we

have about our processes the better we are at improving them. 
Cause-and-effect diagrams are one quality tool that is simple
yet very powerful in helping us better understand our
processes.  They are a tool that fosters team work, helps to
show the true picture, can be used on any issue of the
business, and they are fun. 
1. This example is take from the book Statistical Quality Design and Control. Authors are Richard E. DeVor,
Tsong-how Chang, and John W. Sutherland.

Check Sheets
Purpose:

Check sheets allow the user to collect data from a process in

an easy, systematic, and organized manner.
Data Collection:

Before we can talk about check sheets we need to understand

what we mean by data collection.  Process improvement
actions are always based on information obtained from data
collected from the actual process.  This collected data needs to
be accurate and relevant to the quality problem being
analyzed if we wish for our information about this problem to
also be accurate.  Information is based on data.  There are
three primary steps that need to be taken before any data can
be collected.  The first is to establish a purpose for collecting
this data.  This is based on the quality problem that is going to
be investigated.  Second, we need to define the type of data
that is going to be collected.  Data can be collected in two
ways: Measurable data such as length, size, weight,
time,...etc., and countable data such as the number of defects. 
Which type of data to use again depends on the quality
problem being investigated.  The third step is to determine
who is going to collect that data and when it should be
collected.  Usually one can use statistical guides on when to
take samples, or data points, from a process.  As for who is
going to collect the data, what is important is that the person
collecting the data understands the purpose of collecting this
data and his role in the data collecting process.
Check Sheets:

Check sheets are some of the most common tools used for
collecting data.  They allow the data to be collected in an
easy, systematic, and organized manner.  Also, data collected
using check sheets can be used as input data for other quality
tools such as Pareto diagrams. There are four main types of
check sheets used for data collection (custom check sheets can
also be designed to fit specific needs):
1.Defective Item Check Sheet:

This type of check sheet is used to identify what types of

problems or defects are occurring in the process.  Usually
these check sheets will have a list of the defects or problems
that may occur in the process.  When each sample is taken, a
mark is placed in the appropriate column whenever a defect or
a problem has been identified.  The type of data used in the
defective item check sheets is countable data.  Table 1 below
shows an example of a defective item check sheet for the
wave solder manufacturing process.
Table 1. Wave Solder Defect Count.

Defect Insufficient Solder Excessive

Cold Solder Blow  Holes
Type Solder Bridge Solder
Frequency xxxxxxx xx xxx xxxxxxxxxxxxxx xx
Total 7 2 3 14 2

2. Defective Location Check Sheet:

These type of check sheets are used to identify the location of

the defect on the product.  They are used when the external
appearance of the product is important.  Usually this type of
check sheet consists of a picture of the product. On this
picture, marks can be made to indicate were defects are
occurring on the surface of the product. 
3. Defective Cause Check Sheet: 
This type of check sheet tries to identify causes of a problem
or a defect.  More than one variable is monitored when
collecting data for this type of check sheets.  For example, we
could be collecting data about the type of machine, operator,
date, and time on the same check sheet.  Table 2 below is an
example of this type of check sheets.  As we can see most of
the error is occurring at machine 2 and at the afternoon shift. 
This could suggest that machine 2 has problems when it is run
in the afternoon shift.

Table 2. Defect cause check sheet.

Machine 1 Machine 2
Operator A Morning X X
 Afternoon XX XXXXXX
Operator B Morning X XX
Afternoon XX XXXXXXXXX
X= Number of times the supervisor is called per day.

4. Checkup Confirmation Check Sheet:

This type of check sheet is used to ensure that proper
procedures are being followed. These check sheets usually
will have a list of tasks that need to be accomplished before
the action can be taken.  Examples of checkup confirmation
check sheets are final inspection, machine maintenance,
operation checks, and service performance check sheets.  
Conclusion: 
Check sheets are helpful tools for proper data collection. 
They are easy to use and allow the user to collect data in a
systematic and organized manner.  Many types of check
sheets are available.  The most common are the defective
item, defective location, defective cause, and checkup
confirmation check sheets.

Control Charts
Purpose:

To ensure that the process is in control and to monitor process

variation on a continuous basis.
Statistical Process Control:

The development and use of statistics and statistical theories

about distributions and how they vary has become the corner
stone of process improvement.  More known as statistical
process control, SPC, this method allows the user to
continuously monitor, analyze, and control the process.  It is
based on the understanding of variation and how it effects the
output of any process.  Variation is the amount of deviation
from a design nominal value.  Not every product that is
produced will exactly match it's design nominal values. 
That's why we have tolerances on the nominal values to judge
whether a product is acceptable or not.  But the closer we are
to the nominal value the better the product is.  Control charts
is one SPC tool that enables us to monitor and control process
variation.
Common and Special Cause Variation:

Variation in a process can be caused by, or related to, two

types of causes.  They are common or system causes and
special or local causes.  Special causes are problems that arise
in a periodic fashion.  They are somewhat unpredictable and
can be dealt with at the machine or operator level.  Examples
of special causes are operator error, broken tools, and machine
setting drift.  But this type of variation is not critical and only
represents of small fraction of the variation found in the
process.
Common causes are problems inherent in the system itself. 
They are always present and effect the output of the process. 
Examples of common causes of variation are poor training,
inappropriate production methods, and poor workstation
design.  As we can see, common causes of variation are more
critical on the manufacturing process than special causes.  In
fact Dr. Deming suggests that about 80 to 85% of all the
problems encountered in production processes are caused by
common causes, while only 15 to 20% are caused by special
causes.
Control charts:
Developed in the mid 1920's by Walter Shewhart of Bell labs,
this SPC tool has become a major contributor to the quality
improvement process.  It allows the use to monitor and
control process variation.  It also allow the user to make the
proper corrective actions to eliminate the sources of variation. 
Even though they require the user to have some statistical
background, the are relatively easy to construct.  There are
two basic types of control charts, the average and range
control charts.  The first deals with how close the process is to
the nominal design value, while the range chart indicates the
amount of spread or variability around the nominal design
value.  A control chart has basically three line: the upper
control limit UCL, the center line CL, and the lower control
limit LCL.  These lines are computed from samples taken
from the production line.  Each sample represents a point on
the control chart.  A minimum of 25 points is required for a
control chart to be accurate.  To better understand the use of a
control chart and how it is constructed we present the
following example:
Construction of a Control Chart:

A manufacturing process produces gasket material blanks.  In

order to better understand the manufacturing process, it was
suggested to the manufacturing engineer to construct a control
chart.  After meeting with his team, he decides to construct a
control chart for the thickness of the gasket material blanks. 
The team decides to take a sample size of n=4 every have
hour and measure the thickness of each gasket material blank. 
The average of the four thicknesses is then measured as well
as the range for the four parts.  Table 1 below has a list of the
X-bar, or average, of each sample and it's range.
The average= Sum of the thickness for each measured part/
number of parts. 
The range= The biggest value in the sample - the lowest value
in the sample.
Table 1. The X-bar And Range of Each Sample.
8:30a 9:30a 10:30a 11:30a 12:30p
  8am 9am 10am 11am 12pm 1pm
m m m m m
X-
1.25 0.95 0.8 1.0 1.15 0.95 0.95 1.2 1.05 1.0 0.9
bar
rang
0.5 0.45 0.55 0.00 0.25 0.5 0.4 0.65 0.1 0.5 0.35
e
1:30p 2:30p 3:30p 4:30p 5:30p 6:30p
  2pm 3pm 4pm 5pm 6pm
m m m m m m
X-
1.15 0.85 1.0 1.25 0.95 1.05 1.0 1.3 1.15 1.0 1.05
bar
rang
0.45 0.5 0.6 0.5 0.55 0.3 0.5 0.35 0.55 0.4 0.4
e

The next step is to construct the three lines of each of the

average and range control charts:
For the Range chart: 
The center line = The average of all the sample ranges (R-
bar)= 0.425. 
UCL= D4*(R-bar)= 2.283*0.425=0.97
LCL=D3*(R-bar)= Zero *0.425= Zero.

For X-bar or average chart: 

The center line (X-double bar)= is the average of all the
averages of the 25 sample points.= 1.043
UCL= X-double bar + A2*(R-bar).
= 1.043 + (0.729*0.425)=1.486
LCL = X-double bar -A2(R-bar) 
= 1.05 - (0.729*0.425)=0.60
Where the constants D3, D4, and A2 are based on the sample
size of each sample. Any text on the subject will have a table
for these values with respect to the sample size n. Now we are
ready to draw the two control charts: The X-bar control chart:

X-bar Chart for Thickness

-
-
---------------------------------------------UCL=1.486
S 1.400+
a - +
m -+ +
p - +
l - + + + =
e 1.050+-----------------+-----------------------+-+X=1.043
- + + ++ + + ++ +
M - +
e - +
a - +
n 0.700+
---------------------------------------------LCL=0.5999
-
-
-
+---------+---------+---------+---------+---------+
0 5 10 15 20 25
Sample Number

The range control chart:

R Chart for thickness

1.50+
-
S -
a -
m -
p 1.00+
l --------------------------------------------UCL=0.97
e -
-
R - + +
a 0.50+ + + + + + ++ + + _
n ----+---------+---------+-----------------+-+R=0.0.425
g - + + +
e - +
 - +
0.00+-------+------------------------------------LCL=0.000
-

+---------+---------+---------+---------+---------+
0 5 10 15 20 25
Sample Number

Interpretation of the X-bar and Range charts: Variation

reveals it self
through recognizable patterns on the control charts.  A process
is said to be under the influence of common causes only if all
the data points lie within the upper and lower control limits. 
If a point falls out side the control limits then the process is
said to be out of control or under the influence of special
causes.  Remember that special causes are sporadic and in
most cases we can eliminate what is causing them.  Process
improvement can only start with the process being in control. 
Then we look at the control charts to see if any pattern or
trend exists in the graph.  Control charts can detect several
basic patterns that can occur in a process.  These patterns are:
Runs:  A series of consecutive points on the control chart that fall on one side of
the central Line. This indicates that the mean or average of the process has
shifted.
Trends:  A series of points continues to rise or fall in one direction.  This is an
indication that an abnormal condition is operating within the process.  An
example of causes of trends is tool or equipment wear.
Cycles:  A series of points displaying a similar or repeatable pattern.
Jumps:  Occurs when a large shift has occurred between two consecutive points.
Hugging:  A series of points are very close to the central line or the two control
limits.