A forecast method for software failure time based on linear regression
Yongqiang Zhang, Jingjie Yin · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2006
In this paper multivariate regression approach is adopted to predict the software failure time. Firstly, the linear relationship among several adjacent software failure times is analyzed, and construct multivariate linear regression model by mathematical modeling theory. Secondly, the regression coefficient in the multivariate linear regression equation is derived from the method of least square, and then the equation is obtained. Furthermore, the significance test is performed, including F test statistics and t test statistic. And the linear regression equation is amended by the method of amplifying or diminishing the corresponding coefficients according to t test statistic. Finally, it is certified by a simulation example.