No abstract

Abstract-Good practices in software project management are basic requirements for companies to stay in the market, because the effective project management leads to improvements in product quality and cost reduction. Fundamental measurements are the prediction of size, effort, resources, cost and time spent in the software development process. In this paper, predictive Artificial Neural Network (ANN) and Regression based models are investigated, aiming at establishing simple estimation methods alternatives. The results presented in this paper compare the performance of both methods and show that artificial neural networks are effective in effort estimation.

Software effort estimates is an important part of software development work and provides essential input to project feasibility analyses, bidding, budgeting and planning. Analogy-based estimates models emerge as a promising approach, with comparable accuracy to arithmetic methods, and it is potentially easier to understand and apply. Studies show all the models are sensitive to the quality and availability data, thus requiring a systematic data treatment. In this paper, it is proposed a data pre-processing method for use in software effort estimate. The results of it on applying on applying Case Based Reasoning - CBR that enables us to enhance the precision of the estimates.

Produce quality software inside expected time and low cost has been one of the main challenges in the software industry today. Therefore, it is fundamental make estimates of size, effort, resources, cost and time spent in the software development process. Predictive models such as models based on analogy can be an alternative, especially in small and medium sized software development, they need perform reliable estimates. In this paper, we propose a methodology for pre-processing of data for use in software effort estimation. The results of use this methodology, applying Case Based Reasoning -CBR, indicate that the preprocessing enables to improve the accuracy of estimates.

Estimativas precisas em gerenciamento de projetos é um fator crítico. Medidas de tamanho, esforço, recursos, custo, e tempo despendidos no desenvolvimento de software são fundamentais. Valores subestimados de esforço podem fazer com que pressões de tempo comprometam todo o desenvolvimento funcional e até mesmo o teste de software. Por outro lado, valores superestimados podem resultar em projetos não competitivos. Neste artigo, modelos como redes neurais e regressão, são apontados como alternativas para aqueles que não acreditam em modelos de estimativas. Os resultados apresentados comparam o desempenho desses métodos e indicam que estas técnicas são competitivas com os métodos APF, SLIM e COCOMO.