Hello,

Recently I used an algorithm in the open-source statistical program R to produce a decision tree version of a Patient Reported Outcome Measurement questionnaire. This decision tree significantly reduces the amount of questions needed for an accurate score. Therefore, I want to implement the conditions of the decision tree in the Lime survey questionnaire. This can of course be done manually.

However there are 202 different, unique ways of walking down the decision tree, which would mean 202 difference scenarios in Lime survey. Is there any way of formatting the output of the R-algorithm to be implemented directly into Lime survey?

As attachments I have added a visual representation of part of the decision tree and the output of the algorithm. All black boxes are end nodes.



Each string of output is coding for the conditions for an end node. The string contains the conditions for all steps that have to be taken to end up in a certain node and the value of the node.



The conditions are coded like this:
pv1-5 = question ID
%in% = "when question is"
c("0-10") = answers on the question needed to proceed to the next node.
& = separates different levels of nodes.
:number = value of the predicted score

Thank you in advance for your replies

Mark

You might create the survey via a TSV and then import it into Limesurvey.
TSV supports relevance equations. Conditions are not supported.
www.limesurvey.org/manual/Excel_Survey_S..._of_Survey_Structure

I mannaged to export to a .txt and implement my relevance equations to the .txt format. However i can't get it imported into Limesurvey. It only lists "error" when I try to import the .txt file. Does anyone have an idea what might cause this?

LimeSurvey version and build number
Activate debug to 2 to have more information
And always better to use LibreOffice to do tab separated value

thanks for your quick response.

I used version 2.05, but it seems 2.06 does not give the error and just imports the file. So I will update to 2.06