Early Software Reliability Prediction

Building software that is dependable, meets its quality targets, and stays within schedule and budget is hard. Finding faults early in the process is one partial answer, since fixing defects sooner costs far less than fixing them later. This book sets out an early reliability prediction model meant to track a system's reliability at every stage, from gathering requirements through to testing.

The author explores several approaches to this problem. A central one is a method for sorting software modules into two groups: those likely to be fault-prone and those unlikely to be. The book also covers ways to estimate how many faults a system contains overall, and how many should remain at the close of each development phase.

The techniques are presented so that working practitioners can understand, build, and apply them without difficulty. Readers are shown how these estimates reveal more about both their development process and the reliability of the finished product, helping them decide how to allocate their resources more effectively.