Assessing Software Defects using Nano-Patterns Detection

Defects in software systems directly impact a product’s quality and overall customer satisfaction. Assessing defective code for the purpose of locating vulnerable areas and improving software quality and reliability is important for sustained software development efforts. Over the years, various techniques have been used to determine the likelihood that code fragments contain defects, such as identifying code smells, but these techniques have drawbacks. There is a need for better approaches. This thesis assesses software defects using nano-patterns by demonstrating that certain categories of nano-patterns are more defect-prone than others. We studied three open source systems from the Apache Software Foundation and found that ObjectCreator, FieldReader, TypeManipulator, Looping, Exceptions, LocalReader, and LocalWriter nano-patters are more defect-prone than others. Apart from assessing software defects, we expect this new finding will contribute to further research on other applications of nano-patterns and improve coding practices