Software Quality Engineering CS410

1
Software Quality EngineeringCS410

• Class 5
• Seven Basic Quality Tools

2
Seven Basic Quality Tools

• Promoted by Ishikawa (1989)
• Set of statistical tools for quality control
• Can be used to analyze software metrics
• Most applicable to Project Managers
• Statistical process control for SW development is
  more challenging than traditional manufacturing
• Tools can have a positive influence on quality
  improvement and decision making

3
Seven Basic Quality Tools

• Seven Tools
• 1. Checklist (or Check Sheet)
• 2. Pareto Diagram
• 3. Histogram
• 4. Scatter Diagram
• 5. Run Chart
• 6. Control Chart
• 7. Cause-Effect Diagram
• Figure 5.1 p. 130

4
Checklist

• A paper (or softcopy) form used for collecting
  and arranging data so that it can be easily used
  later
• Also a method for ensuring that key points and
  tasks are addressed, or performed, at the correct
  times
• Also a method to assist in knowing when a
  particular task is complete
• Can be implemented as an expert system to guide
  process and give advise

5
Checklist

• Checklist examples
• Common Error Checklist - developed as part of
  DPP, used at stage kickoff to review common
  errors at that stage
• Process Checklists
• Program Fix Checklists - ensure that all the
  required steps for completing a fix have been
  done (example fig. 5.2 pp132-133)
• Code Complete Checklists - ensure that all coding
  standards and procedures have been followed

6
Pareto Diagram

• Pareto Principle (80/20 rule)
• 80 of the defects are contained in 20 of the
  modules and components
• 20 of a particular defect type cause 80 of the
  customer calls
• Pareto analysis helps identify focus areas that
  cause most of the problems
• Pareto Analysis can be charted as frequencies in
  descending rank order with a cumulative
  percentage line (fig 5.3 p. 134 and fig. 5.4 p.
  136)

7
Histogram

• A graphic representation of frequency
  distribution
• Should convey complete information at a glance
• Examples
• Defect distributions by Sev level (fig. 5.5(A) p.
  137)
• Defect distribution by days open (fig. 5.5(B) p.
  137)
• Profile of Customer Satisfaction (fig. 5.6 p.
  137)

8
Run Charts

• A real-time statement of quality and/or workload.
• Defect arrival rates weekly
• Defect backlog
• Delinquent fixes
• Should include target for comparison
• Can track significant events, and show results
• Figure 5.7 p. 138

9
Run Charts

• Good tool for project and schedule management.
• Run Chart can display the typical S Curve
• Tracks actual vs. planned over time
• Completion of design review
• Completion of code inspection
• Completion of code integration
• Completion of component test
• Completion of system test
• etc.

10
Scatter Diagrams

• Usually relates to investigative work (I.e.
  investigating relationships)
• Requires precise data
• Works well for correlation analysis
• For example Code complexity vs. Defect Rate (fig
  5.9 p. 140)
• Shows positive correlation
• Can be used as a predictor
• Can be used as a tool to reduce defect rate (by
  reducing code complexity)

11
Scatter Diagrams

• Can also be used for identifying/prioritizing
  areas requiring action items
• For example Defects rates for reused components
  in multiple projects (fig. 5.10 p. 141, and fig.
  5.11 p. 142)
• Helps identify chronic problem components
• Helps focus action items on greatest payback

12
Control Chart

• A tool used in Statistical Process Control (SPC)
• Can be used real-time for improving consistency
  and stability
• A tool for measuring Process Capability
• Process Capability - inherent variation in the
  process in relation to the specification limits
• The smaller the process variation the better the
  processs capability
• Process Capability is a difficult thing to
  estimate in software projects, however Control
  Charts are a useful quality management tool.

13
Control Chart

• Types of control charts
• X-bar - sample averages
• S chart - standard deviations
• Median chart - individuals
• P chart - proportion nonconforming
• NP chart - number nonconforming
• C chart - number nonconformities
• U chart - nonconformities per unit
• Most appropriate to SW Engineering
• P chart - when percentages are involved
• U chart - when defect rates are involved

14
Control Chart

• Control limits must be calculated and included on
  the chart
• Different ways of calculating control limits
  exists.
• Historical data
• Weighted averages
• Standard Deviations
• SW metrics that can be control charted
• Defects per component/KLOC (fig. 5.12 p. 146)
• Phase effectiveness (fig. 5.14 p. 147)
• Backlog Management Index

15
Cause-Effect Diagram

• Also known as a Fishbone Diagram
• Not widely used in SW development
• Graphically relates causes and effects
• Effect can be a positive or negative
  characteristic
• Examples
• Register Allocation Defects (negative) fig. 5.15
  p. 148
• Effective Inspections (positive) fig 5.16 p. 149

16
Seven Basic Quality Tools Summary

• Statistical tools (scientific methods) are used
  heavily in manufacturing, and are slowly being
  incorporated into SW engineering.
• Some tools can be used together effectively
• Pareto Diagrams, Cause-Effect Diagrams, and
  Scatter Diagrams, can be used together to
  identify dominant problems and their root causes.
• Control Charts can be used to monitor process
  stability.

17
Seven Basic Quality Tools Summary

• Keys to a tools value
• Consistent use
• Pervasive use
• Automation
• Integration
• As a tool for process improvement

18
Seven Basic Quality Tools Summary

• Ishikawas Tools a.k.a. Seven Old Tools
• Other tools - Seven New Tools
• Affinity Diagram
• Relations Diagram
• Tree Diagram
• Metric Chart
• Matrix Data Analysis Chart
• Process Decision Program Chart
• Arrow Diagram