Enhancing Software Effort Estimation with Ant Colony Optimization Algorithm and Fuzzy-Neural Networks

Software effort estimation remains a persistent challenge and requires serious attention in the early stages of software project management. Inherent uncertainties arising from incomplete and inaccurate requirements pose a significant barrier to reliable estimations. Despite numerous efforts and various techniques proposed for cost estimation, the pursuit of improving estimation accuracy remains essential. In response to this challenge, this study introduces a novel model that integrates an Adaptive Neuro-Fuzzy Inference System with the Ant Colony Optimization algorithm. The model is further compared with well-known evolutionary algorithms such as Differential Evolution, Genetic Algorithm, Artificial Neural Network, and Particle Swarm Optimization. Applying the proposed model to popular software effort estimation datasets, including Albrecht, Desharnais, and Kemerer, demonstrates its superiority over the mentioned algorithms. The improved estimation provided by this model can assist software project managers in better project cost estimation.