Título: Utilizando Redes Neurais para Estimar o Tamanho de um Software
Autores: Souza, Gustavo Alexandre; Pozo, Aurora Trinidad Ramirez; Vergilio, Silvia Regina
Resumo: Prediction models are fundamental at the early stages of the software development where decisions must be taken without the required information. A typical information that is not available in these stages is the number of lines of code (LOC), that is the most known and used software size metric. LOC estimation is a hard task that includes historical data and empirical studies. Usually, models for LOC estimation are obtained using statistical regression methods. However, the characteristics of the LOC estimation task make this problem specially interesting for the application of neural network techniques. This work explores Neural Networks techniques in this context. Two different data sets were used to obtain two models to estimate LOC using respectively the metrics function points and number of components, as independent variables. The results show some insights about the use of Neural Network for this task.
Código DOI: 10.21528/CBRN2003-020
Artigo em PDF: 6CBRN_020.PDF
Arquivo BibTex: 6CBRN_020.bib