Goodness-of-Fit Statistics for RGA Objects

Computes numerical goodness-of-fit statistics for a fitted Reliability Growth Analysis (RGA) model using the time-transformation approach. For a Crow-AMSAA (Power Law NHPP) model with parameters \(\beta\) and \(\lambda\), the transformed values \(W_i = (t_i / t_n)^{\beta}\) should follow a Uniform(0, 1) distribution if the model fits. The Cramér-von Mises and Kolmogorov-Smirnov statistics are computed against this null distribution.

Usage

gof(x, ...)

# Default S3 method
gof(x, ...)

# S3 method for class 'rga'
gof(x, ...)

Arguments

x

An object for which a goodness-of-fit method is defined.

...

Additional arguments passed to methods.

Value

An object of class gof.

See also

Other goodness-of-fit: ppplot.rga(), print.gof(), qqplot.rga()

Examples

times <- c(5, 10, 15, 20, 25)
failures <- c(1, 2, 1, 3, 2)
fit <- rga(times, failures)
g <- gof(fit)
print(g)
#> Goodness-of-Fit Statistics (Crow-AMSAA)
#> ------------------------------------------ 
#> n (observations): 5
#> 
#> Cramer-von Mises statistic (W^2): 0.02835
#> Kolmogorov-Smirnov statistic (D):  0.12347
#> 
#> Smaller values indicate a better fit.
#> W_i = (t_i / t_n)^beta should be Uniform(0,1) under H0.