2) Collect data on past projects.
3) Re-adjust the model coefficients.
4) Iterate back to stage 2, as the calibration processing should be ongoing.
CALIBRATION
Calibration involves fine tuning a algorithmic cost models to a specific local environment to enhance the capability of accurate predictions. It is particularly necessary as there is no standard productivity level across industry due to the subjective nature of software development projects.
Calibration is accomplished by readjusting model coefficients to reflect the productivity levels found in the local environment. The process involves analysing data about past projects. It is also possible to change ‘cost drivers’ within a model or introduce new ones that will reflect the organisation.
Four stages of calibration can be identified.
1) Ensure that values input into the model are consistent with the model,
for example the LOC counting rules.
2) Collect data on past projects.
3) Re-adjust the model coefficients.
4) Iterate back to stage 2, as the calibration processing should be ongoing.