Assessment of Automatically Built Bayesian Networks in Software Effort Prediction

Software prediction unveils itself as a difficult but important task which can aid the manager on decision making, possibly allowing for time and resources sparing. One of the approaches set forth to perform this task has been the application of machine learning techniques. Bayesian networks are perhaps one of the most promising of these. In this context, we present an assessment of automatic Bayesian Networks (i.e. Bayesian Networks solely based on data) on two public datasets and bring forward a discussion of important procedures like discretization, feature subset selection and the validation approach. We carried out a comparison of Bayesian Networks against mean and median models and also against linear regression with a logarithmic transformation, which has been recently deemed in a comprehensive study as a top performer with regard to accuracy. The results support that automatic Bayesian Networks can be competitive against other techniques, but still need improvements in order to overcome linear regression models accuracy-wise. This study also demonstrates the potential benefits of feature selection for Bayesian Networks modelling.