Proposal to develop enhancements and extensions of formal models for risk assessment in software projects

Abstract

Over the past 40 years limited progress has been made to help practitioners estimate the risk and the required effort necessary to deliver software solutions. Recent developments improve this outlook. Researchers from the Naval Postgraduate School developed a formal model for risk assessment used to estimate software project risk. This model is based on easily obtainable software metrics quantifiable early in the software development process. The previous risk assessment model was developed on data collected from a series of experiments conducted on the Vite'Project simulation. This unique approach provided a starting point towards a proven formal model for risk assessment, one that can be applied early in the software development lifecycle. Approaching software risk estimation has never previously been successfully accomplished in this manner. This research provides definitive evidence that software risk assessment can be conducted early in software development using quantifiable metrics and simple techniques. Previous models have been enhanced based on calibrations against post-mortem projects. These enhancements result from many threads of research; extension of input metrics, increased simulations, simulations calibrated on actual projects, and model development. The research delivers an improved risk assessment model, one that has been validated against thousands of post-mortem projects, with applicability on any software development activity.