On Metrics-Driven Software Process

Metrics can drive software processing, because they found the base of its quantizing management. There are two fundamental requirements in engineering: formal modeling and quantitative modeling. We must emphasize that metrics for software engineering is insufficient in quantification now. In order to improve software and software processing, the factors that affect schedule, cost and quality of software development should be measured. Metrics produces adjustable and iterative motions in software processing. Based on Jaynes´ maximum entropy principle, this paper establishes a model to quantify the factors and introduces distance to compare the metric indicators. The authors propose that the metric estimation tree can be used and the nodes that stand for the software attributes in the tree can be marked with their corresponding evaluation values. Dynamic feedback in the software processing will be combined with AHP (analytic hierarchy process), and the project and process of development will be learned and analyzed entirely, concentratedly and dynamically.