Software Effort Estimation for Object Oriented Systems using Computational Intelligence Techniques

Abstract

Planning a project and estimating effort is crucial before developing any software. newlineSoftware development primarily relies on specification-based project planning because newlinesoftware effort assessment is so important for projects to be completed in a timely and costeffective newlinemanner. Due to rapid technological advancements, the need to deploy complex newlinesoftware systems at a cheaper cost and maintain improved software quality is a big issue in newlinethe software business. It has been evident in the literature that various schemes and newlineanalytical methodologies are suggested by the different researchers have planned numerous newlinemethods for effort estimating. Even if each method has its pros and cons, they have all newlinegained significant popularity and practical application. The management of software newlineexpenses clearly presents substantial challenges, particularly on neglect of many real-world newlinecircumstances. The proposed research effort intends to provide a robust analytical newlineframework to develop efficient algorithms for effort estimation in order to be accurate to newlinesoftware effort under unpredictably real-world settings. newlineThis research presents an efficient analytical methodology that aims to accurately newlineestimate efforts required for software cost projection during the overall development stage. newlineThe proposed methodology for building an effort estimation scheme is accomplished with newlinetwo distinct computational frameworks. The modelling of the first framework leverages an newlineapplication of computational intelligence techniques retaining adaptiveness and robustness newlinein its predictive decision towards effort estimation. In this phase, the system model newlineleverages the power of data analytics function to prepare precise and reliable software effort newlinefeatures needed to train the learning model. In addition, an advanced augmented learning newlinemechanisms is adopted to make the proposed learning model adaptive to the training newlinefeatures enabling optimal adjustments of learnable parameter such weights and bias.