What inspired this wish list request?
Using the Life Distribution with process data that is randomly censored (left, right, or interval) and typically, but not always, well-represented by a Weibull distribution. The Life Distribution platform offers no means by which to objectively assess whether the data is in fact well-represented by the selected fit distribution(s). It provides AICC, BIC, and loglikelihood values by which to judge the relative merit of selected models on the dataset, but no test value that could be used, for example, to assess whether the fit is "good enough" to draw meaningful conclusions from the fitted parameters.
The Distribution platform provides calculation of Anderson Darling statistic and p-value as well as plotting Lilliefors confidence intervals on a Q-Q plot for visual assessment. However, Distribution is not compatible with censored data.
What is the improvement you would like to see?
Provide an option to calculated Anderson Darling statistic and p-value for censored data in the Life Distribution platform as well as option to add prediction confidence interval to the main graph output by Life Distribution.
Why is this idea important?
The requested improvements would provide objective means by which to assess how well the selected model(s) fit the data and would be consistent with the assessment methods included in Distribution.