![](bgb.png)
construction of network and local optimization since
GA is a globally optimal solution).
By the validation of continual versions, the
learned BN method is particularly valuable for the
quality evaluation of the OO systems developed with
the highly-iterative or agile strategy.
There are several threats to validity. The first
threat is that only version series of one software
product (Rhino) were used to train and test the
model. But the paper’s focus is on examination of
the learners in agile process software (not
generalization of the method to general software
modules). We have examined our models across
other different software products and statistically
demonstrated our approach’s advantages in a
previous study (Li and Leung, 2011). The second
threat is that selection of different predictive factors
for different models may damage the validity of the
models. But learning was conducted to get the
optimal prediction performance. Using the same
methods of feature selection and optimal learning
algorithms for different models, the prediction
performance of the models could be comparable no
matter what different predictors were used.
In the future, we will explore the following aspects:
Using additional benchmark datasets (Basili et
al., 1996; Menzies et al., 2007; Pai and Dugan,
2007) from public domain, we will conduct more
empirical validation of the BN in comparison
with other models for the fault-proneness
prediction. This can determine the superiority
and stability of BN for the quality assessment of
agile software.
Given the many data mining and optimization
algorithms, we will explore the effects of
different algorithms on the prediction.
We will investigate the applicability of BN for
the prediction of other aspects (e.g. reliability) of
software quality, using additional metrics (e.g.
slice-based cohesion and coupling) and
qualitative factors.
ACKNOWLEDGEMENTS
This research is partly supported by the Hong Kong
CERG grant PolyU5225/08E, NSFC grant
1171344/D010703, MOST grants (2012CB955503
and 2011AA120305–1).
REFERENCES
Ambler, S. W., R., J., 2002. Agile Modeling: Effective
Practices for Extreme Programming and the Unified
Process. John Wiley & Sons.
Basili, V. R., Briand, L. C., Melo, W. L., 1996. A
validation of object-oriented design metrics as quality
indicators. IEEE Transactions on Software
Engineering 22, 751-761.
Bouckaert, R. R., 1995. Bayesian Belief Network: from
Construction to Inference.
Boyd, N., 2007. Rhino home page.
Briand, L. C., Wust, J., Daly, J. W., Porter, D. V., 2000.
Exploring the relationships between design measures
and software quality in object-oriented systems.
Journal of Systems and Software 51, 245-273.
Bugzilla, D., 2005. Mozilla Foundation.
Cardoso, J., 2006. Process Control-flow Complexity
Metric: An Empirical Validation, IEEE International
Conference on Services Computing (IEEE SCC 06).
IEEE Computer Society, Chicago, pp. 167-173.
Chidamber, S. R., Kemerer, C. F., 1994. A metrics suite
for object-oriented design IEEE Transactions on
Software Engineering 20, 476-493.
Cohn, C., 2006 Agile Alliance Home Page.
CYVIS, 2007. CYVIS.
D'Ambros, M., Lanza, M., Robbes, R., 2012. Evaluating
defect prediction approaches: a benchmark and an
extensive comparison. Empirical Software
Engineering 17, 531-577.
Dirk, V. P., Bart, L., 2004. Customer Attrition Analysis
for Financial Services Using Proportional Hazard
Models. European Journal of Operational Research
157, 196-127.
Elomaa, T., Rousu, J., 1996. Finding optimal multi-splits
for numerical attributes in decision tree learning,
ESPRIT Working Group, NeuroCOLT Technical
Report Series, pp. 1-16.
Fenton, N., Neil, M., Marsh, W., Hearty, P., Radlinski, L.,
Krause, P., 2008. On the effectiveness of early life
cycle defect prediction with Bayesian nets Empirical
Software Engineering 13, 499-537.
Gokhale, S. S., Lyn, M. R., 1997. Regression tree
modeling for the prediction of software quality, Proc.
Of Third ISSAT Intl. Conference on Reliability,
Anaheim, CA, pp. 31-36.
Guo, L., Ma, Y., Cukic, B., Singh, H., 2004. Robust
prediction of faultproneness by random forests, the
15th International Symposium on Software Reliability
Engineering. IEEE Computer Society, Washington,
DC, pp. 417- 428.
Harrison, R., Counsell, S., Nithi, R., 1998. An Evaluation
of the MOOD Set of Object Oriented Software Metrics.
IEEE Transaction on Software Engineering 24, 150-
157.
Heeger, D., 1998. Signal Detection Theory.
Herbsleb, J. D., 2001. Global software development. IEEE
Software 18, 16-20.
Hosmer, D., Lemeshow, S., 2000. Applied Logistic
Regression, 2 ed. John Wiley and Sons.
PredictingFault-pronenessofObject-OrientedSystemDevelopedwithAgileProcessusingLearnedBayesianNetwork
15