Enhancing Software Fault Prediction With Machine Learning by Ekbal Rashid

Author:Ekbal Rashid
Language: eng
Format: epub
Publisher: IGI Global

ANALYSIS AND RESULTS

The main objective of this prediction system is to help development of good quality software. The aim of the chapter is also to explore how case-based reasoning technique can support decision-making and help control dissimilarity in software fault activities, thus ultimately enhancing software quality. Estimating reliability of upcoming software versions based on fault history the fault estimation to enhance test stage effectiveness; task of resources to fix faults, and distinguishing faulty software modules from non-faulty ones. In this chapter, five distance functions used, i.e., Euclidean method, Canberra method, Clark method, Exponential method and Manhattan method, using these five distance functions how machine learning technique like case-based reasoning helps to predict exact matching case from the knowledge base. In this research, mean magnitude of relative error is calculated with the help of dependent variable i.e., development time. It was observed that when I was using the five distance function for the same data set, the results are coming quite good when I was using exponential distance. And a result is derived where acceptable range is within 10%. Figure 11, 13, 15, 17, and 19 shows the accuracy comparisons of different distance functions. As the number of exact matching case increases, the accuracy also increases. Therefore, I can say that exponential method is best for error prediction while using CBR technique. In this work, I display the error relative to the size metric retrieved from the knowledgebase (E1) and size metric of the user (E2). It can be seen in the form of figure (s) (See Figure 5 through Figure 19). Figures 20-22 shows the display before, during and after the Duplicate Data Set has been removed from KBS.

CONCLUSION

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.