Chaudhari, Bimal
2010-May-10 22:13 UTC
[R] predict() without generating the model within R
Is there a predict method/syntax which I could use to generate predictions (and other output from predict() methods) if I have the model parameter estimates from a training dataset but not the data used to generate the original model (the models were generated by a collaborator using STATA and for IRB reasons I am not allowed independent access to the original data)? I have the new/testing data upon which I wish to base my predictions. Thanks, Bimal PS-yes, I know that if my collaborator just used R, he could send me just the model as an Rdata object. That is not going to happen in this instance. Bimal P Chaudhari, MPH Boston University MD Candidate, 2011 Washington University in St. Louis MS Candidate, 2010 Doris Duke Clinical Research Fellow 314-286-2864
On 10/05/2010 6:13 PM, Chaudhari, Bimal wrote:> Is there a predict method/syntax which I could use to generate > predictions (and other output from predict() methods) if I have the > model parameter estimates from a training dataset but not the data used > to generate the original model (the models were generated by a > collaborator using STATA and for IRB reasons I am not allowed > independent access to the original data)? I have the new/testing data > upon which I wish to base my predictions. >That depends on what model is being fit, but in general I'd say it would be impractical to put together enough information for predict() to work properly. However, you can do predictions for many models just from their definition, and you may have enough information to calculate errors as well. But you'll need to program the formulas yourself. Duncan Murdoch> Thanks, > Bimal > > PS-yes, I know that if my collaborator just used R, he could send me > just the model as an Rdata object. That is not going to happen in this > instance. > > > Bimal P Chaudhari, MPH > Boston University > MD Candidate, 2011 > Washington University in St. Louis > MS Candidate, 2010 > Doris Duke Clinical Research Fellow > 314-286-2864 > > ______________________________________________ > R-help at r-project.org mailing list > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide http://www.R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code. >