🚀 go-pugleaf

in software reliability models, it is often assumed that
the data t_1, t_2, ..., t_N are independent with
t_i having df F_i (of a known parametric form, e.g. the t_i's may be considered
the order statistics of ind. exp. r.v.'s  with parameter (N-i+1)*C).

bev littlewood in several papers has suggested that 
  U_i = F_i(T_i; \beta_{i-1}) 
be calculated (as a goodness of fit indicator) 
where beta_{i-1} is the MLE of the parameter beta based on the first i - 1 data.
a plot of the cdf of the U_i's is called the u-plot.  

i was wondering if anybody could give me additional references on the u-plot (whether
within or without software reliability).

please e-mail responses (to albert@ams.sunysb.edu).  thanks.


jose albert
ams department
suny stony brook
stony brook ny 11794-3600