Expert Systems with Applications: An International Journal
| Hi-index | 0.00 |
Prediction models are fundamental in the earlystages of the software development when many times,decisions must be taken without the required information.A typical information that is not available inthese stages is software size metrics, such as lines ofcode (LOC). Models for LOC estimation are obtainedfrom historical data and statistical regression methodsare usually applied. These characteristics make thisestimation problem especially interesting for the applicationof machine learning techniques. To explore thisfact, this work applies Genetic Programming and NeuralNetworks techniques for LOC estimation. Two differentdata sets were used to obtain two models usingrespectively the metrics function points and number ofcomponents. The models are analysed and the machinelearning techniques are compared.