What are metrics used for?
Metrics are intended to make a process more transparent
by measuring and calculating appropriate variables.
This ensures that
- the status of a process can be depicted and
monitored
- weaknesses can be identified and eliminated in
the future
- process sequences can be continuously improved
thanks to the findings of metrics evaluations.
How are metrics encountered in software
development?
Software metrics are generally based on primary
data which can be interpreted (e.g. number
of errors, review effort in hours, etc.). This data is
then generally used to obtain parameters
(e.g. error identification rate, agreement with
planned effort) calculated according to fixed formulae.
What definitions are important in this
context?
A metrics system is defined by the following:
- Measurement time points
- Primary data to be recorded
- Parameters to be derived from this data
- Evaluation criteria, e.g. comparison with
specifications (e.g. permissible range)
- Measures which are to be initiated as a function
of the evaluation criteria
- Tools for supporting recording (e.g. PROCON-PDE)
and evaluation (e.g. PROCON-QDA).
When is the use of a metrics system expedient?
Use of a metrics system is expedient only if a
development methodis employed and the development process
is structured into individual phases. Only then is it
possible to obtain, evaluate and analyze specific metrics
data at predefined points in time.
The tool used in PSE is PROCON (see entry and monitoring
of data in PROCON).
A further type of metrics is static code
analysis
Static code analysis deals with the structure of the
source code of software programs. The analysis results
provide a picture of the size and complexity of the
analyzed programs. A full description can be found in
theme Static code analyses.
Information and support when using metrics is
available from the Support Center for
Effort Estimation and Metrics.
|