Experimental study on feature selection methods for software fault detection

Software fault detection is the process of analyzing the software for identifying the errors before it is being deployed to the customer. The classifier is employed to perform the software fault detection. Therefore, the accuracy of the software fault detection highly depends on the classifier which is employed in fault detection. Developing the classifier with irrelevant and redundant features of the error-prone data deteriorates the accuracy in software fault detect. Therefore, the feature selection process is employed to remove the redundant and irrelevant features from the error-prone data to improve the accuracy in the software fault detection. Hence, this paper presents an experimental study on the performance of the feature selection methods namely gain ratio (GR), Info gain (IG), OneR, ReliefF, and symmetric uncertainty (SU) to develop the highly accurate classifier for improving the accuracy in software fault detection.