A hybrid approach for software cost estimation using polynomial neural networks and intuitionistic fuzzy sets

Software cost estimation (SCE) is an important and critical activity of any software development organisation. It helps the project managers to effectively manage their projects and prevent them from over budgeting. In this study we introduce a new design methodology for software cost estimation using polynomial neural networks (PNNs) and intuitionistic fuzzy sets which resulted in improved SCEs. The performance of the proposed model is tested through a series of experiments on three publicly available software development data, i.e., COCOMO81, NASA93, and Maxwell datasets. The proposed technique of using IFCM (intuitionistic fuzzy C Means) along with PNNs has drastically improved the cost estimations in comparison with the use of fuzzy C means (FCM) with PNN as reported in the literature.