02-09-2017, 03:21 PM
There has been a tremendous growth in demand for software quality over recent ages. As a consequence, the issues are related to the increasingly critical tests. The ability to measure software defects can be extremely important to minimize cost and improve the overall effectiveness of the testing process. The most failures in a software system are found in some of its components. Although there is a great variety in the definition of software quality, it is truly accepted that a project with many defects lacks the quality of the software. Knowing the causes of possible defects as well as identifying the areas of general software process that may need attention from the initialization of a project could save you money, time and work effort. The possibility of early estimation of the probable lack of software could assist in the planning, control and execution of software development activities. An inexpensive method for defect analysis is to learn from past mistakes to avoid futures. Nowadays, there are several sets of data that can be extracted in order to discover useful knowledge about defects.
Different methods of data mining have been proposed for the analysis of defects in the past, but few of them manage to deal successfully with all the previous issues. Estimates of regression models are difficult to interpret and also provide the exact number of failures that are too risky, especially at the start of a project when little information is available. On the other hand, classification models that predict possible failures may be specific, but not so useful to give an idea of the actual number of failures. Many researchers used many techniques with different data sets that predict failures. But there are so many classification rule algorithms that can be effective in predicting failures. All these questions motivate our research in these fields of software failure / prediction of defects.
Different methods of data mining have been proposed for the analysis of defects in the past, but few of them manage to deal successfully with all the previous issues. Estimates of regression models are difficult to interpret and also provide the exact number of failures that are too risky, especially at the start of a project when little information is available. On the other hand, classification models that predict possible failures may be specific, but not so useful to give an idea of the actual number of failures. Many researchers used many techniques with different data sets that predict failures. But there are so many classification rule algorithms that can be effective in predicting failures. All these questions motivate our research in these fields of software failure / prediction of defects.