Modeling Incremental Faults of Software Testing Process Using AR Models

Software reliability characterizes the foremost quality attribute for the software. Being able to predict the number of software faults helps significantly in specifying/computing the software release day, manage project resources which include people and money. Most of the growth models, known in history, count on two or three model parameters to be estimate. In this paper, we explore the use of linear Auto-Regression (AR) model to predict the accumulated faults resides in a software. This model will count on historical measured faults to product the future faults. The developed models will be tested on three types of datasets with excellent results