Information Technology Reference
In-Depth Information
4.2.2.7
Phase Breakdown
As can be seen, the TPA calculation is done at the project level. So from TPA we
have effort estimate and number of hours required for planning, controlling, and
doing testing. But effort estimates for different phases of the project cannot be
derived from this calculation. To get effort and number of hours of work required
for different phases of the project, we need to divide this figure to get hours and
effort required in each phase. There is no well-established formula for doing so.
The test manager can make an estimate and then divide the total effort into these
phases.
4.2.3
Application of TPA
TPA is a somewhat objective method of calculating effort estimation. But it is very
complicated. It requires elaborate sets of data on which complex calculations are
done to arrive at the solution. Not many people use TPA with this many detailed
calculations. At the majority of places, test managers use approximate values for
TPA. This is because all the needed data may not be available to the test manager.
Furthermore, doing all these complex calculations takes a lot of time.
Some aspects of TPA cannot be claimed to be truly objective. For instance, pro-
ductivity measurement is done based on historical data. Even size and test strategy
measurements cannot be said to be truly objective, contrary to claims.
TPA is widely used for providing some initial effort estimate at the bidding stage.
4.2.4
TPA at Bidding Stage
Service providers need a good tool to provide an effort estimate for a project at the
bidding stage. The effort estimate is one of the most important criteria for any ser-
vice provider to win a bid. At the same time, if he underestimates his bid substan-
tially from what could be the actual effort during execution, then he runs the risk
of losing money. For this reason he must have a good tool for a rough initial effort
estimate. TPA comes in handy in this respect.
Until i l d e it a i le d f u nc it ion a l is p e c i i c a it ion is a r e o bt a i ne d , how e ve r, it i is not p o is is ible it o
determine factors such as complexity, interfacing, and so on. Nevertheless, a rough
function point analysis can be performed on the basis of very general specifications.
If a rough function point count is available, a rough TPA can be estimated.
For a rough TPA, a single function is defined whose size is determined by the
total function point count. All function-dependent factors (quality level, usage
intensity, complexity, interfacing, and uniformity) are usually assigned a normal
value. A TPA can then be carried out as described in previous paragraphs.
