An application of Bayesian network for predicting object-oriented software maintainability
read more
Citations
Software engineering economics
A systematic review of software maintainability prediction and metrics
Predicting object-oriented software maintainability using multivariate adaptive regression splines
A Bayesian belief network for IT implementation decision support
Application of TreeNet in Predicting Object-Oriented Software Maintainability: A Comparative Study
References
Classification and Regression Trees.
Classification and regression trees
Software engineering economics
A metrics suite for object oriented design
Bayesian networks and decision graphs
Related Papers (5)
Predicting object-oriented software maintainability using multivariate adaptive regression splines
Frequently Asked Questions (10)
Q2. What are the future works mentioned in the paper "An application of bayesian network for predicting object-oriented software maintainability" ?
Those findings have also confirmed that Bayesian network is indeed a useful modelling technique for software maintainability prediction, although further studies are required to realize the full potential as well as the limitation. This provides an interesting 16 direction for future studies. The results in this paper also suggest that the prediction accuracy of the Bayesian network model may vary depending on the characteristics of dataset and/or the prediction accuracy measure used.
Q3. What measures are commonly used in the software effort prediction literature?
The Bayesian network model’s prediction accuracy is evaluated using some accu-2racy measures, which are commonly found in the software effort prediction literature [16,24].
Q4. How does the network predict the posterior probability distribution of CHANGE?
After the batch learning, the network predicts the posterior probability distribution of CHANGE for each case in the corresponding test subset, by computing the joint probability distribution.
Q5. Why is the Med.Ab.Res. chosen as a measure of the central tendency?
The Med.Ab.Res. is chosen to be a measure of the central tendency because the residual distribution is usually skewed in software datasets.
Q6. How many classes were chosen by random sampling without replacement?
Approximately a two-third of the cases in each dataset is chosen by random sampling without replacement using a function provided in a statistical software package, SPSS 11.0.
Q7. What is the prediction accuracy for the UIMS dataset?
For the UIMS dataset, the Bayesian network model has achieved significantly better prediction accuracy than the regression tree model and the multiple linear regression models.
Q8. Why is the Bayesian network model able to predict uncertainty?
This is due to the ability to explicitly represent uncertainty using probabilities, the ability to incorporate existing human expert’s knowledge into empirical data, and the ability to update the model when new information becomes available.
Q9. What are the results of the Wilcoxon signed-rank tests of the MRE values?
The Wilcoxon signed-rank tests of the MRE values have also confirmed strong evidence that the Bayesian network model’s MMRE value is significantly lower and thus, better than those of the other models.
Q10. What is the definition of a Bayesian network?
From this point of view, Bayesian networks can be considered as a network of events connected by the probabilistic dependencies between them.