An approach for OO software size estimation using Predictive Object Point Metrics

Abstract

For estimating software, system size is the main parameter of the system development effort. It affects substantially on accurate estimation of effort of development. The Predictive Object Point (POPs) input gives an estimate of the size of the software for which the estimation is required. POPs are a metric suitable for estimating the size of object oriented software, based on the behaviors that each class is delivering to the system along with top level inputs describing the structure of a system. However there is no real mapping of Source Lines of Code (SLOC) to POPs exists. This paper is an attempt to map the Predictive Object Point Metrics with software size which may help in further prediction of effort. This may also help in estimation of cost as well as schedule measurement of an OO system. The proposed method of mapping between POP and software size has been empirically investigated. KLOC has been estimated in terms of EKLOC through POP count using the linear regression equation. The results are presented here to show that how POP Count may be mapped to corresponding software size (KLOC) of an object oriented system.