Using differential evolution in the prediction of software effort

Estimation of software is a very important and crucial task in the software development process. Due to the intangible nature of software, it is difficult to predict the effort correctly. There are number of options available to predict the software effort such as algorithmic models, non-algorithmic models etc. Estimation of Analogy has been proved to be most effective method. In this, the estimation is based on the similar projects that have been successfully completed already. If the parameters of the current project, matches well with the past project then it is easy to calculate the effort for current project. The success rate of the effort prediction largely depends on finding the most similar past projects. For finding the most relevant past project in estimation by analogy method, the computational intelligence tools have already been used. The use of Artificial Neural Networks, Genetic Algorithm has not fully solved the problem of selection of relevant projects. The main problems faced are Feature Selection and Similarity Measure between the projects. This can be achieved by using Differential Evolution. This is a population based search strategy. The Differential Evolution is used to compare the key attributes between the two projects. Thus we can get most optimal projects which can be used for the estimation of effort using analogy method.