Improving Case Based Software Effort Estimation by an Ant Colony Algorithm

Abstract

Predicting accurate efforts at the early stages of the Software Life Cycle is one of the greatest challenges in software industry. Effort estimation remains a very difficult task since the software requirements are not well known and understood/or the details have not yet been specified. Thus, most of software effort estimation models are built on using historical project data. For this purpose, CBR based prediction models have been extensively used. In CBR techniques, choosing the most similar case has a strong impact on the prediction accuracy, nevertheless, it remains a hard task especially on the presence of alternatives projects; projects that are equally similar. This paper proposed a new method to select the most appropriate case by using ant colony optimization algorithm. The ACO algorithm will be used to search for the best case of past cases based on the different features similarity values (FR_similarity, NFRs_similarity and DPs_similarity). To verify the efficiency and performance of the proposed model, an example was conducted and the results were compared with that of the real estimation. The results of the example show that proposed model in this paper is an attractive alternative to retrieve the most appropriate case and is useful and beneficial for decision making during the effort estimation.