Data and Cost handling Techniques for Software Quality Prediction Through Clustering

Analysis of Data quality is an important issue which has been addressed as data warehousing, data mining and information systems. It has been agreed that poor data quality will impact the quality of results of analyses and that it will therefore impact on decisions made on the basis of these results. An attempt to improve classification accuracy by pre-clustering did not succeed. However, error rates within clusters from training sets were strongly correlated with error rates within the same clusters on the test sets. This phenomenon could perhaps be used to develop confidence levels for predictions. The main and the common problem that the software industry has to face is the maintenance cost of industrial software systems. One of the main reasons for the high cost of maintenance is the inherent difficulty of understanding software systems that are large, complex, inconsistent and integrated. The main reason behind the above phenomena is because of different size and level of arrangements. Decomposing a software system into smaller, more manageable subsystems can aid the process of understanding it significantly. Different algorithms construct different decompositions. Therefore, it is important to have methods that evaluate the quality of such automatic decompositions. In our paper we present a brief survey on software quality prediction through clustering.