Information Technology Reference
In-Depth Information
until subsequent phases. Difficulty in implementing a poor design may occur
without warning, or problems in testing may be the result of poor test planning
or previously undetected coding defects. CPR analysis can be a good starting
point for identifying problem areas, and problem reports included with CPRs
may provide insight for risk assessments.
Cost-Estimating Relationships (CERs)
: Cost and schedule are estimated by de-
termining effort based on algebraic relationships between a dependent (effort or
cost) variable and independent variables. This method ranges from using a sim-
ple factor, such as cost per LOC on a similar program with similar contractors, to
detailed multivariant regressions based on several similar programs with more
than one causal (independent) variable. Statistical packages are available com-
mercially for developing CERs, and if data are available from several completed
similar programs (which is not often the case), this method may be a worthwhile
investment for current and future cost and schedule estimating tasks. Parametric
model developers incorporate a series of CERs into an automated process by
which parametric inputs determine which CERs are appropriate for the program
at hand.
Of these techniques, the most commonly used is parametric modeling. There is
currently no list of recommended or approved models; however, the team will need
to justify the appropriateness of the specific model or other technique they use. As
mentioned, determining which method is most appropriate is driven by the availability
of data. Regardless of which method used, a thorough understanding of software's
functionality, architecture, and characteristics, and contract is necessary to accurately
estimate required effort, schedule, and cost.
SOFTWARE METRIC PLAN
9
5.8
For measurement to be effective, it must become an integral part of the team decision-
making process. Insights gained from metrics should be merged with process knowl-
edge gathered from other sources in the conduct of daily program activities. It is the
entire measurement process that gives value to decision making, not just the charts
and reports. Without a firm metrics plan, based on issue analysis, you can become
overwhelmed by statistics, charts, graphs, and briefings to the point where the team
has little time for anything other than ingestion.
Not all data are worth collecting and analyzing. Once the team development project
is in-process, and your development team begins to design and produce lines-of-code,
the effort involved in planning and specifying the metrics to be collected, analyzed,
and reported on begins to pay dividends.
9
http://www.stsc.hill.af.mil/resources/tech docs/gsam3/chap13.pdf.
Search WWH ::
Custom Search