Identifying correlations of findings for building process improvement packages using graph clustering

Abstract

AbstractSoftware process improvement (SPI) begins with process assessment based on a process reference model such as CMMI. Process improvement action items in SPI are determined according to the identified strengths and weaknesses of the current practice. Therefore, given that a list of assessment findings has been identified, it is important to analyze correlations of findings and identify relevant findings for building improvement items. However, correlation analysis requires expertise and considerable efforts, which makes it difficult for practitioners to perform it in process improvement projects. In this work, we present a CMMI‐based method for identifying correlations of findings and building improvement packages using graph clustering techniques. We evaluate the method using industrial data. Copyright © 2015 John Wiley & Sons, Ltd.