Publication

Abstract : Fault is considered one of the most common issues in the software life cycle. Some of these errors are not identified during the testing phase, causing problems throughout the production process and execution time. The technology and complexity in the field of software development are increasing day by day, increasing the difficulties in fault prediction prior to the production process. Earlier, the prediction of failure in software is pushed into production by enhancing the system’s reliability and quality. The wastage of time and cost due to failure in runtime can be avoided by implementing a better prediction system. Some traditional methods are available for predicting failures in different aspects and cannot provide proper output. This paper proposes an advanced failure prediction system by using a machine learning (ML) algorithm. It is capable to detect software faults in a very fast and accurate manner. This system is developed by combining different ML techniques for comparing the performance of each technique and it also determines how they influence results. Ensemble learners are used for combining ML algorithms by soft voting and plurality voting for performing early prediction in software module failure. A graphical user interface (GUI) and the dashboard have been developed for obtaining better output representation.