chapter 3 statistical process control tools and...
TRANSCRIPT
43
CHAPTER 3
STATISTICAL PROCESS CONTROL TOOLS AND CMM
This chapter discusses the SPC and the control charts in detail. The applicability
of the SPC and the control chart in software industry are discussed and the selection of
the suitable chart for the analysis also discussed. The concepts of CMM and levels of
CMM are discussed.
3.1 Tools for Statistical Process Control
Continuous quality improvement process assumes and even demands that a team
of experts in the field as well as a company leadership actively use quality tools in their
improvement activities and decision making process. Quality tools can be used in all
phases of production process, from the beginning of the product development up to
product marketing and customer support. At the moment there are a significant number of
quality assurance and quality management tools on disposal to quality experts and
managers, so the selection of most appropriated one is not always an easy task.
The underlying concept of statistical process control is based on a comparison of
what is happening today with what has happened previously. If what is happening today
is minutely different from the earlier, it is still in a state of control. A snapshot of the
process typically performs or builds a model of how the process will perform and
calculate the control limits for the expected measurements of the output of the process.
From the earlier calculations/models built it is basically calculated the control limits and
once the control limits have been stabilized then the process is left alone to be understood
alone in the state of statistical control.
Data is collected from the process and compared to the control limits. The
majority of measurements should fall within the control limits and the measurements that
fall outside the control limits are examined to see if they belong to the same population as
our initial model.
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
44
In general, software development processes have the following characteristics:
1. Software development processes are more heavily dependent on human activities
and require creativity than those of manufacturing processes (Caivano, 2005). It
follows that the measurement results of software processes can easily vary and
hence become instable.
2. Involves multiple common causes related to the first point, there are various
factors that can affect the performance of the processes, e.g., tools, methods, type
of software and development environment as pointed out by Florac and Carleton
(1999). These factors bring about a multiple common cause system. Therefore the
data obtained will be a mixed set of common cause results.
3. Hard to obtain a large set of homogeneous data, especially those data which have
important business value. Unlike manufacturing processes where many parts are
daily produced, software development processes usually result in a relatively
small set of work products. Furthermore each process execution is a creative and
unique activity, which makes it hard to obtain large set of homogeneous data.
Despite all these characteristics, SPC is still important in software processes. SPC
can give guidance to stabilize and improve processes. It can also be used to demonstrate
the improvement effects.
The key to overcome these difficulties is to put more emphasis on processes rather
than products. For example, using not only the product data but also process data can
increase the available data significantly. Under the assumption that the definition of
software process involves people, tools and working environment as suggested by
Carleton and Florac, (1999), SPC can help to stabilize human activities. It also helps to
identify and separate multiple common causes.
There are many ways to implement process control, but the key monitoring and
investigating tools include:
1. Histograms
2. Run charts
3. Pareto charts
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
45
4. Cause and effect diagrams
5. Flow diagrams
6. Scatter diagrams
7. Control charts
3.1.1 Histogram
A histogram is a bar graph usually used to present frequency data. Histograms
provide an easy way to calculate the distribution of data over different categories. The
histogram is an effective, practical working tool in the early stages of data analysis.
For constructing a histogram the data are categorized when large data are
available. Then the data are collected and sorted into the categories. The data are counted
for each category and the diagram is drawn. A sample histogram is shown in Figure 3.1.
Each category finds its place on the x-axis and the bars in the histogram will be as high as
the value of the category.
A histogram may be interpreted by asking three questions:
1. How well is the histogram centered? The centering of the data provides
information on the process aim about some mean or normal value.
2. How wide is the histogram? Looking at the histogram width defines the
variability of the process about the aim.
3. What is the shape of the histogram? The data are expected to form a
normal or bell-shaped curve. Any significant change or anomaly usually
indicates that there is something wrong going in the process, which is
causing the quality problem.
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
46
Figure 3.1 Sample histogram
Histograms are limited in their use due to the random order in which samples are
taken and lack of information about the state of control of the process. Because samples
are collected without regard to order, the time-dependent or time–related trends in the
process are not captured. This lack of information on process control may lead to
incorrect conclusions being drawn and hence, inappropriate decisions being made. Still
with these considerations in mind, the histogram’s simplicity of construction and ease of
use make it an invaluable tool in the elementary stages of data analysis.
3.1.2 Run chart
A Run chart is a simple graphic representation that displays data in the order that
they occur and shows a characteristic of a process over time. It is often known as a line
chart or a line graph outside the quality management field. Run charts allow us to
understand objectively if the changes are made to process or system overtime lead to
improvements and do so with minimal mathematical complexity. These methods of
analyzing and reporting data are of greater value to improvement projects and teams than
traditional aggregate summary statistics that ignore time order. Because of this utility and
simplicity, the run chart has wide potential application in software development for
project managers and decision-makers. Run charts also provide the foundation for more
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
47
sophisticated methods of analysis and learning such as Shewhart (Control) charts and
planned experimentation.
Run charts originated from control charts. Shewhart, (1939) initially designed and
developed a system for bringing processes into statistical control by developing ideas,
which would allow for a system to be controlled using control charts. Run charts evolved
from the development of these control charts, but run charts focus more on time patterns
while a control chart focuses more on acceptable limits of the process.
Run chart is used to understand the trends and shifts in a process or variation over
time or to identify decline or improvement in a process over time. In a run chart, events
shown on the y-axis are plotted against a time period on the x-axis. A sample run chart is
given in Figure 3.2.
Figure 3.2 Sample run chart
Run chart rules
The key rules indicating special cause variation are:
Rule 1: shift
A shift in the process is six or more consecutive points either all above or all
below the median line. Values that fall on the median line do not count towards or break
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
48
Observations
a shift–skip value on the median and continue counting. The shift rule is shown in
Figure 3.3.
Figure 3.3 Run chart shift rule
Rule 2: trend
A trend is five or more consecutive points all going up or all going down. If two
successive points have the same value, only the first counts towards the trend–like values
do not count towards or break a trend. Second point can be ignored and counting can be
continued with the next point on the chart. The rule 2 is shown in Figure 3.4.
Figure 3.4 Run chart trend rule
Rule 2
0
5
10
15
20
25
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Me
as
ure
or
Ch
ara
cte
ris
tic
Observations
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
49
Rule3: runs
A run is a series of points either above or below the median line (points on the
median line are ignored). The number of runs can be easily calculated by counting the
number of times the data line crosses the median line and adding one to that total. A
special table is used to determine the upper and lower limits on this value.
Rule 4: astronomical value
An astronomical value is one that appears highly unusual in a ‘blatantly obvious’
fashion – not just taking the highest or lowest point on the chart. Not a precise definition,
but one that can generally be agreed with colleagues. Figure 3.5 shows the diagram for
applying rule 4.
Figure 3.5 Run chart astronomical value
3.1.3 Pareto chart
Pareto charts are the graphical tool used in Pareto analysis. A Pareto chart is a bar
chart that displays the relative importance of problems in a format that is very easy to
interpret. The Pareto chart is named after Vilfredo Pareto, a 19th
century economist who
postulated that a large share of wealth is owned by a small percentage of the population.
This basic principle translates well into quality problems. A Pareto chart is a series of
bars whose heights reflect the frequency or impact of problems. The bars are arranged in
descending order of height from left to right. The sample Pareto chart is shown in Figure
Rule 4
0
5
10
15
20
25
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Me
as
ure
me
nt
or
Ch
ara
cte
ris
tic
Observations
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
50
3.6. This means the categories represented by the tall bars on the left are relatively more
significant than those on the right. This bar chart is used to separate the “vital few” from
the “trivial many”. These charts are based on the Pareto principle which states that 80
percent of the problems come from 20 percent of the causes. Pareto charts are extremely
useful because they can be used to identify those factors that have the greatest cumulative
effect on the system and thus screen out the less significant factors in an analysis. Ideally,
this allows the user to focus attention on a few important factors in a process.
Check sheets are relatively simple forms used to collect data. They include a list
of nonconformities and a tally of nonconformities. Check sheets should also include the
name of the project for which data are being collected, the shift when the items are
produced, the names of persons collecting the data, dates of data collection and of
production and the location of data collection. Check sheets are not mandatory to
construct Pareto charts.
Figure 3.6 Sample Pareto chart
A Pareto chart is a good tool to use when the investigating process produces data
that are broken down into categories and that counts the number of times each category
occurs. A Pareto diagram puts data in a hierarchical order, which allows the most
significant problems to be corrected first. The Pareto analysis technique is used primarily
to identify and evaluate nonconformities, although it can summarize all types of data. It is
the perhaps the diagram most often used in management presentations.
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
51
Making problem solving decisions isn’t the only use of the Pareto Principle. Since
Pareto charts convey information in a way that enables you to see clearly the choices that
should be made, they can be used to set priorities for many practical applications. Some
examples are:
Process improvement efforts
Skills the division should have
Customer needs
Suppliers
Investment opportunities
3.1.4 Cause and effect diagram
A Cause and effect diagram is a tool that is used to identify, sort and display
possible causes of a specific problem or quality characteristic. The relationship between a
given outcome and all the factors that influence the outcome is graphically illustrated
through this diagram. This diagram is also called as “fishbone diagram “because of the
way it works.
Cause and effect diagram helps the users to identify the possible root cause, the
basic operations, for a specific effect, problem or condition. Using the diagram the team
can take the corrective action by analyzing the existing problems. A cause and effect
diagram is helpful for organizing the known or possible causes of quality or lack of it.
The diagram helps the team in a very systematic way because of its structure.
The advantages of constructing a cause and effect diagram are as follows:
Structured approach is applied to determine the root causes of the problem
or quality characteristic
Encourages group participation and utilizes group knowledge of the
process
Shows possible causes of variation in a process
Improves knowledge of the process by helping the team to learn more
about the factors at work and how they relate
Defines areas where data should be collected for further study
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
52
Constructing a cause and effect diagram is a three step process and discussed below.
Step 1
Writing down the effect to be investigated and draw the 'backbone' arrow to it. In the
example shown below the effect is 'Incorrect deliveries'.
Step 2
Identifying all the broad areas of enquiry in which the causes of the effect being
investigated may lie. For incorrect deliveries the diagram may then become:
Step 3
This step requires the greatest amount of work and imagination form the team to write in
all the detailed possible causes in each of the broad areas of enquiry. Each cause
identified should be fully explored for further more specific causes which, in turn,
contribute to them.
Figure 3.7 Sample cause and effect diagram
This process is continued of branching off into more and more directions until
every possible cause has been identified. The final result shows a sort of a 'mind dump' of
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
53
all the factors relating to the effect being explored and the relationships between them.
Cause and effect diagrams enable a team to focus on the content of the problem. It
focuses the team on causes, not symptoms. It is an effective tool that allows people to
easily see the relationship between factors to study the process, situations and for
planning. The sample cause and effect diagram is shown in Figure 3.7.
Causes in a cause and effect diagram are frequently arranged into the following
four categories
1. Manpower, methods, materials and machinery (recommended for manufacturing)
2. Equipment, methods, materials and people (recommended for administration and
service).
3.1.5 Flow diagram
A process flow chart is simply a tool that graphically shows the inputs, actions
and output of a given system. The purpose of flow chart is to help people to understand
the process. A flow chart illustrates the activities performed and the flow of resources and
information in a process. Two types of flow charts are particularly useful – high level and
detailed. A high level flow chart illustrates how major groups of related activities, often
called “sub processes”, interact in a process. Typically, four to seven sub processes are
shown in a flowchart. By including only basic information, high level flowcharts can
readily show an entire process and its key sub processes.
A detailed flow chart provides a wealth of information about activities at each
step in a sub process. It shows the sequence of the work and includes most or all of the
steps, including rework steps that may be needed to overcome problems in the process. A
quality improvement team can increase the detail to show the individuals performing
each activity or the time required to complete each activity. If necessary, the link between
various points in the sub process and other high level flow charts of the process can also
be shown. An organization pursuing quality improvement is constantly looking for ways
to improve the effectiveness and efficiency of its work.
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
54
“Effectiveness” means producing the required results of output when needed.
“Efficiency” means simply producing those results or outcomes the first time with
minimum resources. In order to generate ideas on how to be more efficient and effective,
it is helpful to define and document how activities are actually performed. Flow charts
are useful for this purpose. Flow charts can be useful to identify activities in a process
that reduce our effectiveness and efficiency. For example, some activities may be
redundant or repeated, other may be unnecessary. Activities may be performed in
sequence, when they could be conducted at the same time to reduce the overall time for
the process. Flow charts can be used to identify conditions that cause delays and
bottlenecks. This can bring focus to problems at various points within the process that
need further evaluation and improvement.
3.1.6 Scatter diagram
A scatter diagram is a tool for analyzing relationships between two variables. One
variable is plotted on the horizontal axis and the other is plotted on the vertical axis. The
pattern of their intersecting points can graphically show the relationship pattern. Most
often scatter diagram is used to prove or disapprove cause and effect relationships. While
the diagram shows relationships, it does not by itself prove that one variable causes the
other. In addition to showing possible cause and effect relationships, a scatter diagram
can show the two variables are from a common cause that is unknown or that one
variable can be used as a surrogate for the other. A sample scatter diagram is shown in
Figure 3.8.
Scatter diagrams can be used to examine theories about cause and effect
relationships and to search for the root causes of an identified problem. Also it can be
used to design a control system to ensure that gains from quality improvement efforts are
maintained.
The following steps are followed when using scatter diagrams:
1. Selecting and defining the two variables to be analyzed in the individual chart.
2. Measuring the two variables, if data has not already been collected.
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
55
3. The set of data pairs should consist of at least thirty, but preferably more like
one hundred data points.
4. Designing the chart by placing the independent variable on the horizontal axis.
The independent variable is the factor believed to be governing relationship
between the two variables.
5. On the vertical axis, inserting the dependent variable that is, the factor
believed to change in proportion to the independent variable.
6. Plotting the data pairs themselves in the chart area.
7. Examining the completed chat, looking for pattern that indicates a connection
between the two variables.
8. If correlation patterns are identified, investigating any third variable
involvement before drawing definite conclusions.
Figure 3.8 Sample scatter diagram
Scatter diagrams generally shows one of six possible correlations between the variables:
Strong Positive Correlation : The value of Y clearly increases as the value of X increases
Strong Negative Correlation : The value of Y clearly decreases as the value of X increases.
Weak Negative Correlation : The value of Y clearly decreases slightly as the value of X
increases.
Complex Correlation : The value of Y seems to be related to the value of X, but the
relationship is not easily determined.
No correlation : There is no demonstrated connection between two variables.
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
56
3.1.7 Control charts
There are two principal types of Quality Assurance: “Sampling inspection of
incoming/outgoing materials (acceptance sampling) and control charts for ongoing
processes (process control). Control charts differentiate between the process being in
control (within an accept range of random variation) and out of control (outside the
acceptable range)” (Antony et al, 2000). A process that is operating with only ‘chance
causes of variation’ present is said to be in statistical control. A process that is operating
in the presence of assignable causes is said to be out of control and the eventual goal of
SPC is the elimination of variability in the process. The control chart can be seen as part
of an objective and disciplined approach enable us to take correct decisions regarding
control of the process, including whether or not to change control parameters. Process
parameters should never be adjusted for a process that is in control as this will result in
degraded process performance (William et al, 2006). The primary purpose of a control
chart is to detect whether a major change or shift is imminent or has occurred in a process
resulting in an alteration of that process. The selection of the appropriate control chart is
vital for the success of the quality of process. The control chart is one of the seven basic
tools of quality control (Tague, 2004).
The statistical argument for control chart relies upon the central limit theorem.
The central limit theorem suggests even if the underlying population from which a series
of observations gathered is not normally distributed, the resulting distribution of averages
from the sample observations will be normally distributed around the average values and
that 99.73 percent of the values will be contained within three sigma deviations of that
average value. Using this theorem, a control chart can be constructed.
A control chart consists of:
Points representing a statistic (e.g., a mean, range, proportion) of measurements of
a quality characteristic in samples taken from the process at different times.
The mean of this statistic using all the samples is calculated (e.g., the mean of the
means, mean of the ranges, mean of the proportions). A center line is drawn at the
value of the mean of the statistic.
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
57
The standard error (e.g., standard deviation/sqrt(n) for the mean) of the statistic is
also calculated using all the samples.
Upper and lower control limits (sometimes called "natural process limits") that
indicate the threshold at which the process output is considered statistically
'unlikely' are drawn typically at 3 standard errors from the center line.
The chart may have other optional features, including:
Upper and lower warning limits, drawn as separate lines, typically two standard
errors above and below the centerline.
Division into zones, with the addition of rules governing frequencies of
observations in each zone.
Annotation with events of interest, as determined by the quality engineer
in-charge of the process's quality control chart.
If the process is in control, all points will plot within the control limits. Any
observations outside the limits or systematic patterns within, suggest the introduction of a
new (and likely unanticipated) source of variation, known as a special cause variation.
Since increased variation means increased quality costs, a control chart "signaling" the
presence of a special cause requires immediate investigation. This makes the control
limits very important decision aids. The control limits indicates about process behavior
and have no intrinsic relationship to any specification targets or engineering tolerance. In
practice, the process mean may not coincide with the specified value of the quality
characteristic.
The purpose of control charts is to allow simple detection of events that are
indicative of actual process change. This simple decision can be difficult where the
process characteristic is continuously varying; the control chart provides statistically
objective criteria of change. When change is detected and considered good its cause
should be identified and possibly become the new way of working, where the change is
bad then its cause should be identified and eliminated. The purpose in adding warning
limits or subdividing the control chart into zones is to provide early notification if
something is incorrect. Instead of immediately launching a process improvement effort to
determine whether special causes are present, the quality engineer may temporarily
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
58
increase the rate at which samples are taken from the process output until it's clear that
the process is truly in control. Note that with three sigma limits, common causes result in
signals less than once out of every forty points for Skewed processes and less than once
out of every two hundred points for normally distributed processes (Wheeler, 2010).
Control charts are categorized into the following two types:
1. Attribute control charts
2. Variable control charts
Attribute control charts are used when collected data or the measurements are
categorized as acceptable or not acceptable. These attributes provide counting data. This
data or actual number or failures or the fraction or percent of failures is charted (Smith,
1991). Variable control charts are applied to data that follow a continuous distribution.
3.1.7.1 Attribute control charts
Stability of the systems can be monitored using attribute control charts. For
constructing attribute charts data count or percentage is accumulated. Attribute charts are
the outcome of an assessment using pass or fail criteria.
Attribute charts consist of:
p-chart (Graph of fraction non-confirming)
np-chart (Graph the number of non-conformities in a sample of n pieces)
u-chart (Draft the defect /n)
c- chart (Graph the defects in a sample of n pieces)
p-charts and np-charts are used to graph non-conformities. c-charts and u-charts
are used to graph non-conforming units. “Non-conformity may be imperfection or the
presence of some non-preferred feature” (Freeman and Mintzas, 1999). A non-
conforming unit on the other hand is defined as, “A non-confirming unit, however, may
fail to meet the assessment of some criteria because of one or more non-conformities”
(Freeman and Mintzas 1999). The accept/reject boundaries must be clearly defined when
attribute control charts are constructed. Also, the accept/reject characteristics should be
agreed with by the customer.
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
59
3.1.7.1.1 p-chart
The p-chart is the most commonly used attribute chart. The p represents the
fraction or percent, of the number of items that are unacceptable (or defective). The p-
chart is most helpful in monitoring and controlling the percentage of defective parts in a
production run (Amsden et al., 1989). It tracks the fraction of nonconforming items in a
sample run. Samples may consist of consecutive items taken at a specified time according
to a random sample plan or 100 percent of the production for a specified time period
(hour, day and week). The acceptable/non-acceptable decision of the process may be
based on one characteristic or several, but a defective (non-acceptable) item is counted as
defective only once, even though it may have several defects (Smith, 1991).
Prior to actual constructing the p-chart, preparation must occur to ensure the
accuracy of the intended data. The process must be clearly defined. Quality control teams
can be effective in developing flow charts and cause and effect diagrams for the process
analysis. The characteristics that will be managed or measured should also be defined.
When individual judgment is used for deciding product conformity, inspectors must all
be consistent and in total agreement as to what exactly is classified as defective (Smith,
1991).
The procedures for creating a p-chart can be split into six steps:
The first is to gather all needed data, selecting the size, frequency and number of
samples. The sample size should be large enough to ensure that most of the samples will
have a nonzero number of defects. The sample size should be big enough to give an
average of five or more defectives per sample. A sample size of at least 50 units is a good
starting point. The calculations and interpretation of the chart are easier when the sample
size is kept constant. When constant sample times are used and the sample size varies, a
single set of control limits can be used as long as the sample sizes do not differ by more
than 25 percent of the average sample size. If the sample sizes differ by more than 25
percent, separate control limit calculations will be needed (Amsden et al., 1989).
Sample frequency should also be a factor in the production schedule. The p-chart
should give an accurate depiction as much as possible of the process at the specific point
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
60
in question. Every item produced in that time should have an equal chance of being
chosen in a sample (Smith, 1991). The number of actual samples should be at least 20.
The data collection timeframe should be long enough to track all possible sources of
variation (Smith, 1991).
The second step in constructing a p-chart is calculating the p (percentage) for each
sample. This is accomplished by dividing the number of defects (np) by the actual sample
size (n) [p = np/n]. If the number of defects in the sample of 50 units were 4, the p would
equal .08 [p = 4/50; .08).
The third step in the p-chart is setting the scale for the chart or graph. After
several p values have been calculated, make the scale from 0 to approximately twice the
largest know p value. This will allow for all the p values to fit on the graph (Smith,
1991).
The fourth step is to plot all p values and connect all plot points with a straight
line (Amsden et al., 1989).
Calculating the p-bar and control limits is the fifth step to the p-chart. The p-bar is
calculated by adding all np’s (total number of defects) and dividing by the total amount
of units in all samples. A solid line should be added to the chart to indicate the p-bar
value.
The Upper Control Limits (UCL) and the Lower Control Limits (LCL) are
calculated using the following formulae
UCL = pbar + 3
LCL = pbar - 3
The last step is the interpretation of the chart. There are three possible situations
to consider. First, all the p values are inside the control limits. If this is the case, this can
be indicative that the process is in statistical control. The inherent variations also will be
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
61
at work with the process. The second situation to consider is that one or two p values are
outside the control limits. The usual practice is to not include these values or any sample
information from these samples. All calculations also need to be refigured including the
p-bars and control limits. The chart then should be re-examined. If any p value is out of
the control limits, the process is not in statistical control. The process and causes need to
be determined and resolved. The p-chart process should begin again to ensure that the
process has been fixed (Amsden et al., 1989).
Figure 3.9 Sample p-chart
If three or more p values are outside the control limits, the process is clearly out-
of-control. causes must be determined immediately to resolve such issues.
Shifts of seven or more points to a higher or lower levels may represent that the
process has been affected somehow or that the exact cause of the defect has somewhat
changed. Trends of a run of seven consecutive points either up or down indicate that
something is causing the defects to change in a gradual manner (Smith, 1991). This type
of pattern indicates that something irregular is affecting the process. The causes could be
anything from poorly trained personnel to inconsistent inspections. If this occurs separate
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
62
p-charts may need to take place at other points in the process. The sample p-chart is
shown in Figure 3.9.
p-charts as well as other factors of SPC have been widely accepted among
“quality practitioners as an aid for monitoring, managing, analyzing and improving
process performance by eliminating causes of variations” (Antony, 2000). The use of this
type of data permits a scientific management style in which decisions are made “based on
facts, rather than guesswork and better products can be produced with less scrap and
rework” (Fine, 1997).
3.1.7.1.2 np-chart
An np-chart is a data analysis technique for determining if a measurement process
has gone out of statistical control. It is sensitive to changes in the number of defective
items in the measurement process. The “np” in np-charts stand for the np (the mean
number of success) of a binomial distribution.
The np control chart consists of:
Vertical axis = the number of defectives for each sub-group
Horizontal axis = the sub-group designation
A sub-group is frequently a time sequence (Example, the number of defectives in
a daily production run where each day is considered a sub-group). Sub-group size must
be constant and also normally large subgroups are needed if the times are equally spaced,
the horizontal axis available can be generated as a sequence.
In addition, horizontal lines are drawn at the mean number of defectives and at the
upper and lower control limits. The distribution of the number of defective items is
assumed to be binomial. This assumption is the basis for the calculating the upper and
lower control limits.
Control limits for the np-chart are calculated on the basis of binomial distribution
and an approximation based on the central limit theorem.
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
63
The control limits are calculated as:
LCL = np - 3
UCL = np + 3
Where n is the number of items and p is the proportion of defective items. Zero
serves as a lower bound on the LCL value. The sample p-chart is shown in Figure 3.10.
Figure 3.10 Sample np-chart
np-charts are used to analyze the results of process improvements. Here it is
consider how the process is running and comparing it how it ran in the past. Finally, np-
charts are used for standardization. Without a control chart, there is no way to know if the
process has changed or to identify sources of process variability.
3.1.7.1.3 u-chart
A u-chart is an attribute control chart used with data collected in subgroups of
varying size. u-charts show how the process is measured by the number of non-
conformities per item or group of items, changes over time. Non-conformities are
defects or occurrences found in the sample subgroup. They can be described as any
characteristic that is present but should not be or any characteristic that is not present but
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
64
should be. For example, a scratch sent, bubble, missing button and a tear are all
nonconformities. u-chart is used to determine if the process is stable and predictable, as
well as to monitor the effects of process improvement theories.
A u-chart is particularly useful when the item is too complex to be ruled as simply
confirming or nonconforming. A u-chart is appropriate when the area of opportunity for a
defect varies from subgroup to subgroup. This can be seen in the shifting of UCL and
LCL lines that depend on the size of the subgroup. When areas of opportunity are
appropriately measured and a Poisson model applies, u-charts are the tool of choice.
Defects per thousand lines of source code, defects per function point and system failures
per day in steady-state operation are all examples of attributes data that are candidates for
u-charts. Defects per module and defects per test, on the other hand, are unlikely
candidates for u-charts, c-charts or any other charts for that matter. These ratios are not
based on equal areas of opportunity and there is no reason to expect them to be constant
across all modules or tests when the process is in statistical control.
By virtue of their dependence on Poisson distributions, both u-charts and c-charts
must satisfy the following conditions (Wheeler and Chambers, 1992):
They must be based on counts of discrete events
The discrete events must occur within some well-defined, finite region of
space, time or product
The events must occur independently of each other
The events must be rare, relative to the opportunity for their occurrence
Although u-charts may be appropriate for studying software defect densities, the
use of Poisson models is not validated through any of the studies for these situations. For
example, if modules are deliberately made small as the tasks that they perform are
inherently difficult to design and program, then simple defect densities are unlikely to
follow the same Poisson process across all modules. When the opportunities for
observing the event are not constant, as when differently sized portions of code are
examined or tested for defects, the counts must be converted into rates -such as defects
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
65
per thousand lines of code or defects per thousand source statements, before they can be
compared. The rates are computed by dividing each count (ci) by its area of opportunity
(ai). The rate that results is denoted by the symbol ui. Thus,
Ui = Ci/ai
Once values for Ui have been calculated, they can be plotted as a running record
on a chart. The sample np-chart is shown in Figure 3.11.
No. of Observations
De
fect
De
nsit
y
1413121110987654321
0.16
0.14
0.12
0.10
0.08
0.06
0.04
0.02
0.00
_U=0.06
UCL=0.1271
LCL=0
3
1
1
HP-Grp Maintenance
Figure 3.11 Sample u-chart
The center line and control limits for u-charts are obtained by finding u, the
average rate over all the areas of opportunity and using these formulas:
∑ Ui
Ubar = --------
∑ ai
LCL = Ubar- 3
UCL = Ubar+ 3
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
66
The presence of the area of opportunity ai in the formulas for the control limits for
u-charts means that the limits will be different for each different area of opportunity.
3.1.7.1.4 c-chart
A c-chart is a data analysis technique for determining if a measurement process
has gone out of statistical control. The c-chart is sensitive to changes in the number of
defective items in the measurement process. The “c” in the control chart stands for
“counts” as in defectives per lot.
A c-chart or count chart, is an attribute control chart that displays how the number
of defects or nonconformities, for a process or system is changing over time. The
number of defects collected for the area of opportunity can be either a group of units or
just one individual unit on which defect counts are performed. The c-chart is an
indicator of the consistency and predictability of the level of defects in the process. The
sample c-chart is shown in Figure 3.12.
When constructing a c-chart, it is important that the area of opportunity for a
defect be constant from a subgroup to subgroup since the chart shows the total number
of defects. When the number of items tested within subgroup changes, then a u-chart
should be used, since it shows the number of defects per unit rather than total defects.
Figure 3.12 Sample c-chart
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
67
3.1.7.2 Variable control charts
Control limits for attributes data are often computed in ways quite different from
control limits for variables data. Variables data (sometimes called measurement data) are
usually measurements of continuous phenomena.
Examples : measurements of length, weight, volume and
speed.
Software examples : elapsed time, effort expended, years of
experience, memory utilization and cost of
rework.
The following are the types of variable control charts:
X-bar chart
R-chart
s-chart
Individual chart
Moving Range chart
X-bar chart is based on the average of a subgroup. Subgroups of 2 to 30 samples
may be used when computing the control limits for the X-bar chart when based on the
range.
R-chart takes into account the range of a subgroup. Subgroup sizes may be as
small as 2 or as large as 30. It is important to construct and interpret an R-chart before the
X-bar chart. When the R chart indicates that process variation is in control, analyze the
X-bar chart is analyzed otherwise X-bar is not meaningful.
s-chart takes into account the standard deviation of a subgroup. There is no limit
to the subgroup size. Process variability can be controlled by either a R-chart or a
standard deviation chart (s chart) depending on how the population standard deviation is
estimated. s-chart is used to determine whether the standard deviation has changed.
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
68
Individual chart displays each value. A subgroup size is used to compute the
limits, with value of 2 being most common, although the subgroup size may be as large
as 30.
3.1.7.2.1 XmR chart
An individual and moving range (XmR) chart is a pair of control charts for
processes with a subgroup size of one. This is used to determine, if a process is stable and
predictable and also it creates a picture of how the system changes over time. The
individual (X) chart displays individual measurements. The moving range (mR) chart
shows variability between one data point and the next. XmR charts reflect data which do
not lend themselves to form subgroups with more than a one measurement. The only
condition that needs to be checked before using the XmR control chart is that the average
count per sample is greater than one.
For example, a process repeats itself frequently or it appears to operate differently
at different times. In such case, grouping the data might mask the effects of such
differences. One can overcome this problem by using an XmR chart whenever there is no
rational basis for grouping the data.
An individual chart is equivalent to X-bar but reported to single observation, not
to samples. Used when the nature of the process is such that it is difficult or impossible to
group measurements into subgroups. This occurs frequently in low volume production
and in situations in which continuously varying quantities within the process are process-
related variables. The solution is to artificially create subgroups from the data and then
calculate the range of each subgroup. This is done by creating rolling groups (most often
pairs) of data through time and using the pairs to determine the range R. The resulting
ranges are called moving ranges.
Control limits in XmR chart are calculated from moving range (mR). A range is
based on the absolute value of consecutive differences in observations. The first step in
calculating control limits is to estimate the average of the moving range.
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
69
Count the number of time periods, n.
Calculate the absolute value of the difference of every consecutive value,
call this moving range.
Add the moving ranges and divide by n-1 to get the average moving range.
The second step in constructing a XMR chart is calculating the UCL. UCL is also
called as Upper Natural Process Limit (UNPL). The upper control limit is average of the
observations plus a constant ‘E’ times the average moving range. The constant ‘E’
depends on how many consecutive observations are included in the moving range. The
value of the correction factor is chosen so that 99% of the data fall within the control
limits.
If the moving range is calculated from 2 consecutive time periods then the
correction factor E is 2.66.
The Upper Control Limit can be calculated as:
UCL = Average of observations + 2.66 * Average of moving range
The third step in constructing a XMR chart is calculating the LCL. LCL is also
called as Lower Natural Process Limit (LNPL). The lower control limit is average of
observation minus 2.66 times the average range.
The Lower Control Limit is calculated as:
LCL = Average of observations – 2.66 * Average of moving range
The last step is the interpretation of the chart. There are four possible situations to
consider. First, if any one of the values of Xi is outside the control limit of UCL. The
second situation to consider is that any of two out of three successive values should be in
2 sigma and 3 - sigma limits. If four out of five successive values of Xi are in 1 sigma
and 2 sigma means the process is totally out of control. The last step for interpreting the
XMR chart is to consider that the eight successive values are on the same side of the
centerline. In mR chart if any point Xi is outside the UCL then the process is said to be
out of control. The sample XmR-chart is shown in Figure 3.13.
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
70
Figure 3.13 Sample XmR chart
3.2 CMM Practices
Capability Maturity Model broadly refers to the process improvement approach
that is based on a process model. CMM also refers specifically to the first such method,
developed by the Software Engineering Institute in the mid -1980’s, as well as the family
of process model that is followed. A process model is a structured collection of practices
that describe the characteristics of effective process, the practices that included are those
proven by experience to be effective.
X CHART- USER REQUIREMENT SPECIFICATION
-80.00
-60.00
-40.00
-20.00
0.00
20.00
40.00
60.00
80.00
100.00
120.00
PROJE
CT1
PROJE
CT2
PROJE
CT3
PROJE
CT4
PROJE
CT5
PROJE
CT6
PROJE
CT7
PROJE
CT8
PROJE
CT9
PROJE
CT1
0
PROJECTS
DA
YS
Xi Effort
Xiavg
LNPLx
ULPNx
Sigmax
2Sigmax
3Sigmax
1-Sigmax
2-Sigmax
3-Sigmax
mr CHART-USER REQUIREMENT SPECIFICATION
0.0010.0020.0030.0040.0050.0060.0070.0080.0090.00
100.00
PROJE
CT1
PROJE
CT2
PROJE
CT3
PROJE
CT4
PROJE
CT5
PROJE
CT6
PROJE
CT7
PROJE
CT8
PROJE
CT9
PROJE
CT1
0
PROJECTS
DA
YS
mR
CLr
UCLr
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
71
CMM can be used to assess an organization against a scale of five process
maturity levels. Each level ranks the organization according to its standardization of
processes in the subject area being assessed. The subject areas can be as diverse as
software engineering, systems engineering, project management, risk management,
system acquisition, information technology service and personnel management.
The CMM is a way to develop and refine an organization’s process. The first
CMM was for the purpose of developing and refining software development processes. A
maturity model is a structural collection of elements that describe characteristics of
effective process. A maturity model provides:
a place to start
the benefit of a community’s prior experiences
a common language and a shared vision
a framework for prioritizing actions
a way to define what improvement means for the organization
A maturity model can be used to benchmark for accessing different organizations
for equivalent comparison. It describes the maturity of the company based upon the
project the company is dealing with and its clients.
In the 1970s, technological improvements made computers more widespread,
flexible and inexpensive. Organizations began to adopt more and more computerized
information systems and the field of software development grew significantly.
Unfortunately, the influx of growth caused growing pains; project failure became more
commonplace not only because the field of computer science was still in its infancy, but
also because projects became more ambitious in scale and complexity. In response,
individuals such as Edward Yourdon and Gerald Weinberg, published articles and books
with research resulted in an attempt to professionalize the software development process.
Humphrey’s, (1989) CMM was described in the book managing the Software
Process. The CMM as conceived by Humphrey is based on the earlier work of Phil
Crosby. Active development of the model by the SEI began in 1986. The CMM was
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
72
originally intended as a tool to evaluate the ability of government contractors to perform
a contracted software project. Though it comes from the area of software development, it
has been continued and can continue to be widely applied as a general model of the
maturity of processes in software (and other) organizations.
The model identifies five levels of process maturity for an organization. Within
each of these maturity levels are s Key Process Areas (KPA) which characterize that level
and for each KPA there are five definitions identified:
1. Goals
2. Commitment
3. Ability
4. Measurement
5. Verification
The KPAs are not necessarily unique to CMM, representing - as they do - the
stages that organizations must go through on the way to becoming mature. The
assessment is supposed to be led by an authorized lead assessor. One way in which
companies are supposed to use the model is first to assess their maturity level and then
form a specific plan to get to the next level. Skipping levels is not allowed.
3.2.1 Levels of the CMM
There are five levels of CMM which are described in details.
Level 1 - Initial
Processes are usually ad hoc and the organization usually does not provide a stable
environment. Success in these organizations depends on the competence and heroics of
the people in the organization and not on the use of proven processes. In spite of this ad
hoc, chaotic environment, maturity level 1 organizations often produce products and
services that work; however, they frequently exceed the budget and schedule of their
projects.
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
73
Organizations are characterized by a tendency to over commit, abandon processes
in the time of crisis and are not able to repeat their past successes again. Software project
success depends on having quality people.
Level 2 - Repeatable
Software development successes are repeatable. The processes may not repeat for
all the projects in the organization. The organization may use some basic project
management to track cost and schedule.
Process discipline helps ensure that existing practices are retained during times of
stress. When these practices are in place, projects are performed and managed according
to their documented plans.
Project status and the delivery of services are visible to management at defined
points (for example, at major milestones and at the completion of major tasks).
Basic project management processes are established to track cost, schedule and
functionality. The minimum process discipline is in place to repeat earlier successes on
projects with similar applications and scope. There is still a significant risk of exceeding
cost and time estimate.
Level 3 - Defined
The organization’s set of standard processes, which is the basis for level 3, is
established and improved over time. These standard processes are used to establish
consistency across the organization. Projects establish their defined processes by the
organization’s set of standard processes according to tailoring guidelines.
The organization’s management establishes process objectives based on the
organization’s set of standard processes and ensures that these objectives are
appropriately addressed.
A critical distinction between level 2 and level 3 is the scope of standards, process
descriptions and procedures. At level 2, the standards, process descriptions and
procedures may be quite different in each specific instance of the process (for example,
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
74
on a particular project). At level 3, the standards, process descriptions and procedures for
a project are tailored from the organization’s set of standard processes to suit a particular
project or organizational unit.
Level 4 - Managed
Using precise measurements, management can effectively control the software
development effort. In particular, management can identify ways to adjust and adapt the
process to particular projects without measurable losses of quality or deviations from
specifications. At this level organization sets a quantitative quality goal for both software
process and software maintenance.
Sub processes are selected that significantly contribute to overall process
performance. These selected sub processes are controlled using statistical and other
quantitative techniques.
A critical distinction between maturity level 3 and maturity level 4 is the
predictability of process performance. At maturity level 4, the performance of processes
is controlled using statistical and other quantitative techniques and is quantitatively
predictable. At maturity level 3, processes are only qualitatively predictable.
Level 5 - Optimizing
Focusing on continually improving process performance is done through both
incremental and innovative technological improvements. Quantitative process-
improvement objectives for the organization are established, is continually revised to
reflect changing business objectives and used as criteria in managing process
improvement. The effects of deployed process improvements are measured and evaluated
against the quantitative process-improvement objectives. Both the defined processes and
the organization’s set of standard processes are targets of measurable improvement
activities.
Process improvements to address common causes of process variation and
measurably improve the organization’s processes are identified, evaluated and deployed.
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
75
Optimizing processes that are nimble, adaptable and innovative depends on the
participation of an empowered workforce aligned with the business values and objectives
of the organization. The organization’s ability to rapidly respond to changes and
opportunities is enhanced by finding ways to accelerate and share learning.
A critical distinction between maturity level 4 and maturity level 5 is the type of
process variation addressed. At maturity level 4, processes are concerned with addressing
special causes of process variation and providing statistical predictability of the results.
Though processes may produce predictable results, the results may be insufficient to
achieve the established objectives. At maturity level 5, processes are concerned with
addressing common causes of process variation and changing the process (that is, shifting
the mean of the process performance) to improve process performance (while
maintaining statistical probability) to achieve the established quantitative process-
improvement objectives.
3.3 Chapter Summary
This chapter discussed the methods of several common statistical process control
charts. The definition of a statistical control chart is that it is a graphical device for
monitoring a measurable characteristic of a process for the purpose of showing whether
the process is operating within its limits of expected variation. A major objective of a
statistical process control chart is “to detect quickly the occurrence of assignable causes
of process shifts so that the process can be investigated and corrective actions undertaken
before many nonconforming units are manufactured”. Also the concepts of CMM and the
practices followed in CMM were discussed. This chapter provides a basis for the
identification of methodology for the selection of a control chart used for SPC analysis in
this research. The next chapter discusses the research methodology used to guide the
research.
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.