Improving software quality using machine learning

Software is an entity that keeps on progressing and endures continuous changes, in order to boost its functionality and maintain its effectiveness. During the development of software, even with advanced planning, well documentation and proper process control, are problems that are countered. These defects influence the quality of software in one way or the other which may result into failure. Therefore, in today's neck to neck competition, it is our requirement to control and minimize these defects in software engineering. Software prediction models are typically used to map the patterns of classes of software that are prone to change. This paper highlights the significant analysis in the area's subject to learn and stimulate the association between the metric specifying the object orientation & the concept of change proneness. This would often lead us to rigorous testing so as to find all kinds of possibilities in the data set. We have two views to be addressed: (1) Parameters quantification that affects the quality, functionality and productivity of the software. (2) Machine learning technologies are used for predicting software Here, the focus of the research paper is to equate and compare all of learning methods corresponding to performance parameter with its statistical method & methodology which would often results enhanced. Data points are the basis for prediction of models.

I. Introduction

Software quality means a degree to which software or a process satisfies or associates with the requisites. Software can be summarized to deal with properties such as Complexity which is cyclic in nature, property of cohesion i.e. number of function properties which meet the needs of the customers and thereby provide product satisfaction. Software structures and attributes are categorized in two groups mainly internal and external. The internal quality attributes are those that can be measured during the diverse levels of the development cycle leading to the software SDLC).