Quantitative studies in software release planning under risk and resource constraints
Authors:
- G. RuheGRG. Ruhe
- D. GreerDGD. Greer
- 3 February 2004
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 262-270
Abstract
Delivering software in an incremental fashionimplicitly reduces many of the risks associatedwith delivering large software projects. However,adopting a process, where requirements aredelivered in releases means decisions have to bemade on which requirements should be deliveredin which release. This paper describes a methodcalled EVOLVE+, based on a genetic algorithmand aimed at the evolutionary planning ofincremental software development. The method isinitially evaluated using a sample project. Theevaluation involves an investigation of the trade-offrelationship between risk and the overallbenefit. The link to empirical research is two-fold:Firstly, our model is based on interaction withindustry and randomly generated data for effortand risk of requirements. The results achieved thisway are the first step for a more comprehensiveevaluation using real-world data. Secondly, we tryto approach uncertainty of data by additionalcomputational effort providing more insight intothe problem solutions: (i) Effort estimates areconsidered to be stochastic variables following agiven probability function; (ii) Instead of offeringjust one solution, the L-best (L1) solutions aredetermined. This provides support in finding themost appropriate solution, reflecting implicitpreferences and constraints of the actual decision-maker.Stability intervals are given to indicate thevalidity of solutions and to allow the problemparameters to be changed without adverselyaffecting the optimality of the solution.Keywords
This publication has 8 references indexed in Scilit:
- Has twenty-five years of empirical software engineering made a difference?2003
- A case study of unanticipated incremental change2003
- Prioritisation of system changes using cost-benefit and risk assessments2003
- Preliminary guidelines for empirical research in software engineeringIEEE Transactions on Software Engineering, 2002
- How good are agile methods?IEEE Software, 2002
- Nature's algorithms [genetic algorithms]IEEE Potentials, 2001
- An evaluation of methods for prioritizing software requirementsInformation and Software Technology, 1998
- On spanning tree problems with multiple objectivesAnnals of Operations Research, 1994