Process performance model for predicting Delivered Defect Density in a software scrum project

Abstract

Agile scrum projects are gaining popularity for software development and delivery, due to its inherent attributes of flexibility in execution and quicker time to market. However, the critics of Agile have been raising questions on the predictability of success of Agile execution as the emphasis given to quantitative management using predictive techniques for an Agile project may not be to the extent of large mission critical projects which uses in-process leading indicators to predict outcomes. One possible solution to this challenge is to use the best practices from industry adopted models like the Capability Maturity Model Integration® for planning and managing Agile software delivery. In this paper, we present a study were a prediction model was developed for use in Agile projects for systematic in-process monitoring and decision making. The purpose of this paper is to propose a predictive model as a driver for continuous improvement of Delivered Defect Density in the context of a Scrum project.Our findings show that using the prediction model, the success of Agile project could be forecasted and controlled during its planning and execution phases as against waiting for the final release to The paper starts with defining the background and context of the study, research methodology and study of existing literature in this area. A brief description of the major knowledge areas considered in the study namely Agile, CMMI and High Maturity practices in CMMI is then presented. An introduction on measures, metrics and prediction models is also provided. The development of the Agile Prediction Model and the application of the model Agile SCRUM projects is addressed as the main theme of the paper which concludes with the advantages, shortcomings and limitations of the model developed. 1. Background of the study In organizations where quality of software is a primary goal for the software projects, it is useful to measure and observe Delivered Defect Density as a lagging indicator signifying the quality of developed software. Though this is a good metric to indicate the goodness of the delivered software, the metric lacks the ability to be controlled, as it is an outcome of the software development process and not an in-process metric. In this context, the need for in-process or leading indicators which will help practitioners get visibility on the outcome is important. In this paper, the processes and sub-processes leading to Delivered Defect Density is