It would be great if Distribution and Quantile Profilers in the Reliability Platform optionally showed Likelihood Intervals instead of Wald Intervals.
As Likelihood Intervals tend to be more precise than Wald Intervals, we prefer Likelihood Intervals in Custom Estimation. When accompanying these estimations by Distribution/Quantile Profilers, however, it can get confusing when the intervals used in graphs don't reflect the custom estimations.
Ideally, Profilers could also optionally show one-sided intervals instead of two-sided ones.
Below is an example created with the JMP Sample Locomotive Data where except for the first 5 Failes all other observations were set to "Censored".
1 Comment