The prediction of software complexity based on complexity requirement using artificial neural network

In the recent years, the productivity of software has grown in size, complexity, and also cost. As that software productivity growth, several problems has been appeared in software project management especially that correlated to complexity. One of complexity factors is requirement. A unit of requirement used as an option to the design phase of product development. The requirement is also a main option in verification process. So the the requirement complexity in this research is used as parameter to predict the software complexity. Because of the data pattern to connect between the requirement and the complexity is complex. So that this paper attempt to make a connectivity model between requirement complexity and prediction complexity of software using artificial neural network method with Levenberg Marquadt and Bayesian Regulation algorithm. So it can be seen comparison of experimental results by using the two algorithms.