A technique for function block counting

Abstract A systematic approach to the estimation of software size early in the development cycle can pay valuable dividends for project planning and management. This paper reports the successful application of a method which combined function block (FB) counting with structured analysis, used linear regression analysis to convert FB counts to function points (FPs), and employed statistical measures of variation to create probability ranges for the estimates. A simple regression model provided significant predictive capability when the model parameters were tuned for the development domain and specific project.