Using sensitivity analysis to create simplified economic models for regression testing
ISSTA '08 Proceedings of the 2008 international symposium on Software testing and analysis
On the effectiveness of early life cycle defect prediction with Bayesian Nets
Empirical Software Engineering
Defect cost flow model: a Bayesian network for predicting defect correction effort
Proceedings of the 6th International Conference on Predictive Models in Software Engineering
A COSMIC-FFP approach to predict web application development effort
Journal of Web Engineering
COCOMO-U: an extension of COCOMO II for cost estimation with uncertainty
SPW/ProSim'06 Proceedings of the 2006 international conference on Software Process Simulation and Modeling
Genetic algorithm for optimizing neural network based software cost estimation
SEMCCO'11 Proceedings of the Second international conference on Swarm, Evolutionary, and Memetic Computing - Volume Part I
| Hi-index | 0.00 |
Software cost estimation techniques predictthe amount of effort required to develop asoftware system. Cost estimates are neededthroughout the software lifecycle to determinefeasibility of software projects and to provide forappropriate allocation or reallocation of availableresources. As the cost of a project depends on thenature and characteristics of the project, theaccuracy of the estimates depends on the amountof reliable information about the product to bedeveloped. At the same time, most costestimation models rely heavily on subjectiveexpert evaluations affected by possibly highdegree of imprecision and uncertainty. To assessthe effect of imprecise evaluations, acomprehensive sensitivity analysis has beenperformed on a major cost estimation model,COCOMO II. Results of this analysis aredescribed and explicated in this paper. To reducerisk of drawing biased conclusions, threedifferent methods for sensitivity analysis havebeen employed: mathematical analysis of theestimating equation, Monte Carlo simulation,and error propagation. The results of the first twomethods are very consistent and confirmexpected highest sensitivity of the model to theimprecision of the size estimate. Errorpropagation allows determination of thecombined impact of imprecision in multipleinputs and it is therefore most valuable from thepractical point of view. The results obtained bythis technique also indicate very strongsensitivity to the imprecision in size estimates. Apossible way to cope with imprecise informationin software cost estimation is indicated in theconcluding part of the paper.