 |
 |
This is almost certainly a naive question, or at least poorly posed. I am struggling to put this problem into a Bayesian framework and thence into WINBUGS. I have a SINGLE machine which exhibits aberrant behaviour from time to time .. this is recorded as 0=Normal, 1=Abnormal (failure). I wish to find the cause or causes of this behavior. I can conduct tests on the machine - there are 8 switches which I can switch to high or low according to some sensible testing plan, plus a covariate (temperature) which I can only observe, not measure. Temperature may or may not be influenced by the settings of the switches. Because the fault is intermittent I wish to continue testing even after an instance of the fault is observed. The absence of a fault on a particular run does not enable us to discard possible causative factors, nor does the presence of a fault necessarily point to a particular factor even if the run on which the fault is detected differs only in the level of one factor from a previous run. The number of runs is not fixed a-priori : this perhaps is of no consequence in the Bayesian data analysis framework, except that the user is likely to terminate testing when she has reasonable certainty about the cause of the fault. Some initial "subjective priors" can certainly be provided.. in the form of a statement of belief about the relative "likelihood" of each of the causes. Now here I am stuck. How do I go about setting this up in a Bayes/WinBugs format? Essentially what I want is continuing feedback (after each run) on the relative "likelihood" of each of the causes (point estimate, and perhaps some distribution about the point estimates). Can anyone help? Thank you John Aitchison ------------------------------------------------------------------- This list is for discussion of modelling issues and the BUGS software. For help with crashes and error messages, first mail [log in to unmask] To mail the BUGS list, mail to [log in to unmask] Before mailing, please check the archive at www.jiscmail.ac.uk/lists/bugs.html Please do not mail attachments to the list. To leave the BUGS list, send LEAVE BUGS to [log in to unmask] If this fails, mail [log in to unmask], NOT the whole list