RStudio and neuralnets gwplot does not work

[Oakwilde]

It works fine in R by itself

[mnwright]

Works here. Could you give a reproducible example?

[Oakwilde]

Here is an example. Thanks. Richard library(car) data(Arrests) attach(Arrests) Released<-ifelse(released=="Yes",1,0) Black<-ifelse(colour=="Black",1,0) Year<-year Age<-age Male<-ifelse(sex=="Male",1,0) Employed<-ifelse(employed=="Yes",1,0) Citizen<-ifelse(citizen=="Yes",1,0) Checks<-checks detach(Arrests) work1<-data.frame(Released,Black,Age,Year,Male,Employed, Citizen,Checks) nn1<-neuralnet(Released~.,data=work1,hidden=c(2,2), linear.output = F,err.fct = "ce", rep=3) print(nn1) # no GWs computed Richard Berk berkr@sas.upenn.edu From: "Marvin N. Wright" <notifications@github.com> Reply-To: bips-hb/neuralnet <reply@reply.github.com> Date: Wednesday, April 3, 2019 at 8:05 AM To: bips-hb/neuralnet <neuralnet@noreply.github.com> Cc: "Berk, Richard A" <berkr@sas.upenn.edu>, Author <author@noreply.github.com> Subject: Re: [bips-hb/neuralnet] RStudio and neuralnets gwplot does not work (#28) Works here. Could you give a reproducible example? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub<#28 (comment)>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AcqgWrMVGpWXhkN68pTNc-ZaDeBzCTEXks5vdJjygaJpZM4cUa3q>.

[Oakwilde]

It works after I scaled all of the predictors. What is strange is that I got weights when I did not scale but no generalized weights. Am I missing something in the documentation? Thanks. Richard Richard Berk berkr@sas.upenn.edu From: "Marvin N. Wright" <notifications@github.com> Reply-To: bips-hb/neuralnet <reply@reply.github.com> Date: Wednesday, April 3, 2019 at 8:05 AM To: bips-hb/neuralnet <neuralnet@noreply.github.com> Cc: "Berk, Richard A" <berkr@sas.upenn.edu>, Author <author@noreply.github.com> Subject: Re: [bips-hb/neuralnet] RStudio and neuralnets gwplot does not work (#28) Works here. Could you give a reproducible example? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub<#28 (comment)>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AcqgWrMVGpWXhkN68pTNc-ZaDeBzCTEXks5vdJjygaJpZM4cUa3q>.

[mnwright]

It seems to be a random issue related to convergence. Might it be that the network didn't converge in the cases where gwplot() did not work? On scaled data it should converge faster, so it would make sense in your case.

[Oakwilde]

Thanks. Yes, that could be it, but I got weights and was able plot the visual of the network. There was also no error message that convergence failed (which I have gotten for other applications). Richard Richard Berk berkr@sas.upenn.edu From: "Marvin N. Wright" <notifications@github.com> Reply-To: bips-hb/neuralnet <reply@reply.github.com> Date: Friday, April 5, 2019 at 9:46 AM To: bips-hb/neuralnet <neuralnet@noreply.github.com> Cc: "Berk, Richard A" <berkr@sas.upenn.edu>, Author <author@noreply.github.com> Subject: Re: [bips-hb/neuralnet] RStudio and neuralnets gwplot does not work (#28) It seems to be a random issue related to convergence. Might it be that the network didn't converge in the cases where gwplot() did not work? On scaled data it should converge faster, so it would make sense in your case. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub<#28 (comment)>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AcqgWuWosw3wraZEoQeqhM2YMqGko7MCks5vd1OTgaJpZM4cUa3q>.