An estimate of necessary effort in the development of software projects

Abstract

The estimated of the effort in the development of software projects has already been studied in the field of software engineering. For this purpose different ways of measurement such as Unes of code and function points, generally addressed to relate software size with project cost (effort) have been used. In this work we are presenting a research project that deals with this field, us'mg machine learning techniques to predict the software project cost. Several public set of data are used. The analysed sets of data only relate the effort invested in the development of software projects and the size of the resultant code. For this reason, we can say that the data used are poor. Despite that, the results obtained are good, because they improve the ones obtained in previous analyses. In order to get results closer to reality we should find data sets of a bigger size that take into account more variables, thus offering more possibilities to obtain solutions in a more efficient way.