Error when introducing evidence in an influence diagram

Issue #326 resolved
Francisco Javier Díez created an issue

When I open this network, compile it, introduce first the finding "Result of test = positive" and then "Result of test = negative", an incompatible-evidence message appears. In version org.openmarkov.full-0.3.0-20180110.113538-10.jar (Jan 2018) it worked properly; in version org.openmarkov.full-0.3.0-20180205.095516-23.jar (Feb 2018) the error was there. Apparently it has to do to the extension of evidence: when I compile the network OpenMarkov adds (even in the interface) the finding "Do test? = yes", which derives from the optimal policy, and when I introduce the finding "Result of test = positive" the inference algorithm adds the finding "Therapy = yes", which clashes with the last finding, "Result of test = negative".

Comments (10)

  1. Manuel Luque

    I don't really understand the issue. Aren't "Result of test = positive" and "Result of test = negative" incompatible?

  2. Iago París

    The same is happening to me. This net has by default Do test? with the finding yes in inference mode. In edition mode, however, it doesn't let me Remove finding and if I try to add Do test?=no, OpenMarkov prompts me "Incompatible evidence".

  3. Iago París

    Section 4.5 of the tutorial, step 4 (evaluation) of the second list of steps fail because of this issue

  4. Manuel Luque

    I don’t fully understand this issue. I would like to ask @Francisco Javier Díez and @Iago París to reproduce this issue with the new development version of OpenMarkov (since I solved a related problem this morning).

  5. Jorge Pérez

    I think it is working properly for me. After compiling it (propagate), I introduce the finding "Resultado del test = positive” and, after that "Resultado del test = negative", but no exception appears. I attach a video fragment trying to reproduce this issue.

    Only when I set “Quimioterapia = yes” and later try to modify “Resultado del test” it throws the IncompatibleEvidenceException, but this is the expected behaviour.

  6. Francisco Javier Díez reporter

    I have downloaded the latest version of the source code and it seems to work perfectly. Thank you, Manolo and Jorge.

  7. Log in to comment