Requirements Volatility and Defect Density

Author(s): Yashwant K. Malaiya, Jason Denton
Venue: 10th International Symposium on Software Reliability Engineering
Date: 1999

Type of Experiement: Case Study


The paper discusses the influence of requirement changes taking place during different times by examining the consequences of code replacement, additions, removals, and modifications. The paper compares the defect density with requirement volatility and the defect density with requirements that have not been changed. The two assumptions made are that the testing time duration is fixed and that the number of defects associated with new code is proportional to its size. The paper concludes that if the requirements changes take place close to the release date, the greater the impact it is on defect density. Therefore, the closer the initial requirements are to being complete and correct at the beginning of the development, the greater the reliability of the resulting system.

Process Outline
There are two different ways presented to estimate defect density for a project:

  1. Record the time, type, and extent of requirement changes and use the information to project the defect density at the release time
  2. A model with two parameters - equivalent time and number of equivalent defects injected at that time