Metrics

Report
Metrics
Project and Process Metrics
Why do we measure?
•
•
•
•
•
Assessing project status
Allows us to track risks
Before they go critical
Adjust workflow
See if the team can control the quality of
artifacts
Example metric
• Can you calculate a metric that records the
number of ‘e’ that appear in a program
– Should you?
Effective Metrics
•
•
•
•
•
•
Simple and computable
Empirically and intuitively persuasive
Consistent and objective
Consistent in use of units and dimensions
Programming language independent
Should be actionable
Example metric
• If I calculate the number of defects per
developer and rank them, then using that rank
assign salary raises based on that.
Metrics Guidelines
• Don’t use a single metric to the exclusion of
others
• Metrics should not be used to appraise or
threaten individuals
– Negative metric should be viewed as an
opportunity for improvement
Process vs Project Metrics
• Process across multiple projects
– i.e. Waterfall vs Agile
– Quality: correctness, maintainability, usability,
integrity
– Productivity: earned value analysis
– Statistical SQA data: MTTF, MTTR
– Defect removal efficiency (later)
– Reuse data: percent of components
Defect Removal Efficiency
DRE = E /(E + D)
• E is the number of errors found before
delivery of the software to the end-user
• D is the number of defects found after
delivery.
DRE Example
DRE = E /(E + D)
Defects found during phase:
Requirements (10)
Design (20)
Construction
Implementation (5)
Unit Testing (50)
Testing
Integration Testing (100)
System Testing (250)
Acceptance Testing (5)
By Customer (10)
10 / (10 + 20) = 33%
What are the rest?
Size Oriented Metrics
•
•
•
•
•
•
•
•
errors per KLOC (thousand lines of code)
defects per KLOC
$ per LOC
pages of documentation per KLOC
errors per person-month
Errors per review hour
LOC per person-month
$ per page of documentation
LOC is dangerous
• A Function Point is an estimate of a unit of
utility/productivity
• LOC does not generalize across multiple
languages
Programming
Language
LOC per Function point
avg.
median
low
high
Ada
154
-
104
205
Assembler
C
C++
337
162
66
315
109
53
91
33
29
694
704
178
77
77
14
400
53
63
67
31
41
19
37
77
42
22
11
33
10
7
-
75
263
105
49
55
110
42
16
158
COBOL
Java
JavaScript
Perl
PL/1
Powerbuilder
SAS
Smalltalk
SQL
Visual Basic
63
58
60
78
32
40
26
40
47
Other problems with LOC
• At IBM in the 70s or 80s (I don’t remember)
they paid people per line-of-code they wrote
Object Oriented Metrics
• Number of scenario scripts (use-cases)
• Number of support classes (required to
implement the system but are not immediately
related to the problem domain)
• Average number of support classes per key class
(analysis class)
• Number of subsystems (an aggregation of classes
that support a function that is visible to the enduser of a system)
Questions/Review
• What are some reasons NOT to use lines of code
to measure size?
• What do you expect the DRE rate will be for the
implementation (or construction) phase of the
software lifecycle?
• What about for testing?
• Give an example of a usability metric?
• According to the chart, Smalltalk is much more
efficient than Java and C++. Why don’t we use it
for everything?
Code Metrics
• Static
– Cyclomatic complexity
– Readability
– etc.
• Dynamic
– CPU performance
– Memory usage
– Threading
Cyclomatic Complexity
• A software metric used to measure the complexity of
software
• Developed by Thomas McCabe
• Described (informally)
as the number of
simple decision
points + 1
What is cyclomatic complexity?
• Essentially the number of linearly
independent paths through the code
– The code has no decision statements: complexity
=1
– The code has an if statement, there are two paths
through the code: complexity = 2
• Used as a basis for path testing
Example
• What is the complexity:
public void howComplex(int i) {
while (i<10) {
i++;
System.out.printf("i is %d", i);
if (i%2 == 0) {
System.out.println("even");
} else {
System.out.println("odd");
}
}
}
When do we care?
A number of industry studies have indicated
that the higher V(G), the higher the probability
or errors.
modules
V(G)
modules in this range are
more error prone
Tools will automatically calculate
• NCSS = Non Commenting Source Statements
• CCN = cyclomatic complexity number

similar documents