Minor bug syn.survctree

[edbonneville]

Hello,

Thanks for the nice package! When trying to use synthpop with survival outcomes, I noticed the call would always end up in the environment browser. To fix this, you would need to comment our the (unnecessary?) browser() call in syn.survctree here.

I would also suggest to update the documentation for syn.survctree: currently it states

NOTE that when the function is called by setting elements of method in syn() to "survctree", the parameter minbucket can be supplied to syn() as e.g. survctree.minbucket.

.. which might suggest that the yevent argument should be supplied to syn() as survctree.yevent = event_indicator. However, source code instead suggests that a named list should be supplied as syn(..., event = list("time_variable" = "event_indicator"). Would be nice for the users to know this :-)

[gillian-raab]

Dear Edward, many thanks for this. I wrote that method ages ago and I don't think it has been much used. I am in the process of fixing a couple of other bugs so I will try to do this soon. I was also thinking of writing a method for parametric survival analysis - lognormal (really easy) and Weibull (probably not too bad either. I'd be interested to know what you are using synthpop for. Best wishes Gillian Gillian M Raab Research Fellow Administrative Data Research Centre - Scotland Edinburgh +44 7748 678 551 Home page : https://www.geos.ed.ac.uk/homes/graab

…

________________________________ From: Edouard Bonneville <notifications@github.com> Sent: 11 January 2021 10:41 To: bnowok/synthpop <synthpop@noreply.github.com> Cc: Subscribed <subscribed@noreply.github.com> Subject: [bnowok/synthpop] Minor bug syn.survctree (#14) This email was sent to you by someone outside the University. You should only click on links or attachments if you are certain that the email is genuine and the content is safe. Hello, Thanks for the nice package! When trying to use synthpop with survival outcomes, I noticed the call would always end up in the environment browser. To fix this, you would need to comment our the (unnecessary?) browser() call in syn.survctree here<https://github.com/bnowok/synthpop/blob/e6caaf9d43bca954136b867ff50b7f55d6d9862c/R/functions.syn.r#L606>. I would also suggest to update the documentation for syn.survctree: currently it states NOTE that when the function is called by setting elements of method in syn() to "survctree", the parameter minbucket can be supplied to syn() as e.g. survctree.minbucket. .. which might suggest that the yevent argument should be supplied to syn() as survctree.yevent = event_indicator. However, source code instead suggests that a named list should be supplied as syn(..., event = list("time_variable" = "event_indicator"). Would be nice for the users to know this :-) — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub<#14>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AE3HB7EHL6YDIOWCPN3JJF3SZLIXDANCNFSM4V5LU45A>. The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336.

[edbonneville]

Dear Gillian,

Great! I would agree that parametric survival methods would be welcome.

I have a dataset which I am not allowed to share but for which I wanted to try and make a synthetic copy (for later sharing/code-checking) using synthpop. It has a time to event outcome, and missing baseline covariates which were imputed using multiple imputation. With syn() on the original dataset pre-imputation, the marginal distributions and missing data patterns seem really nicely conserved, but the outcome model (i.e. complete case analysis) results do differ a fair amount - maybe to be expected since this is a single synthetic dataset.

I wonder if it would be better to create the synthetic dataset based on a imputed dataset instead, and then induce the missing values. I will play around with it some more..

Best wishes,
Ed

[bnowok]

synthpop 1.7-0 that is available now on GitHub has an improved error messages and help file for syn.survctree(). The reported bug has been fixed. Parametric survival methods are not available yet.