Quality Management System (QMS) is composed of manuals, policies, procedures, guidelines, templates checklists etc. QMS is defined at the organizational level. It is usually termed as Organizational Set of Standard Process (OSSP). Referring the common OSSP projects defines or composes the process specific to its mode of operation. The OSSP includes a metric repository which is of quantitative in nature, compared to other qualitative things in the repository. A project can compose its process qualitatively without an elaborate quantitative management. In CMMI terms this is about defining the process at CMMI ML 3. And the defined process is known as Project Defined Process (PDP).
Now what about CMMI ML 5? In this quantitative level organization has got baselines and prediction models. So projects use this quantitative data to compose their process. This is known as process composition in CMMI ML 5 terms. Different alternatives for critical processes are identified. The quantitative data associated with these process parameters are measured and their associations with the Process Performance Objectives (PPO) are analyzed. Regression techniques are normally adopted for these analyses. Trade-offs are identified and then the best method suited to project requirements are selected from the alternative methods. For example a projects PPO is to reduce the SLA for an activity ‘X’. Let the critical processes are ‘á’, ‘b’ and ‘c’. There found to be three alternative methods for performing ‘á’, ‘b’ and ‘c’. Regression techniques are used to deduce the relation between ‘á’, ‘b’ and ‘c’ and PPO. Based on cost benefit analysis and probability analysis the optimum method is chosen from the three alternatives for each process. In this way process is composed quantitatively.