zhuwang46
commented
1 year ago Please attach a reproducible data.
On Tue, Mar 28, 2023, 7:35 AM piodag @.***> wrote:
Hi,
I'm fitting a medium sized metabolomics data set called DDA_zip made of 176 rows and 13385 columns with zipath. The following code is working well (I have no interest in the zero part of the regression, the inflation comes from the data collecting procedure)
fit <- zipath(Bites ~ . | 1 , data = DDA_zip, family = "poisson", nlambda=100, lambda.zero.min.ratio=0.005,standardize=T)
As soon as I try to fit a subsample (no replacement) on a new data.frame created with lines selected with sample() from the original df, like shown below, I get the following error
dd.t<-sort(sample(1:nrow(DDA_zip),size=nrow(DDA_zip)*0.8)) DDA_train<-DDA_zip[dd.t,]
fit.r <- zipath(Bites ~ . | 1 , data = DDA_train, family = "poisson", nlambda=100, lambda.zero.min.ratio=0.005,standardize=T)
Errore in zipath_fit(X, Z, Y, weights = weights, offsetx = offsetx, offsetz = offsetz, : NA/NaN/Inf in chiamata a funzione esterna (arg 1)
A similar error is also produced by the cv.zipath() function. Thus I'm not able to CV to find a safer lambda.
Unfortunately, I'm not able to reproduce the error with the bioChemist data set.
I really cannot get the point on this error. Bug? Something related to the data set?
Have you an idea?
Best regards
Giorgio
— Reply to this email directly, view it on GitHub https://github.com/zhuwang46/mpath/issues/5, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALGS6SC52RYWWQE64UP4IXTW6LLIXANCNFSM6AAAAAAWKQFK7M . You are receiving this because you are subscribed to this thread.Message ID: @.***>
piodag
Hi,
I'm fitting a medium sized metabolomics data set called DDA_zip made of 176 rows and 13385 columns with zipath. The following code is working well (I have no interest in the zero part of the regression, the inflation comes from the data collecting procedure)
fit <- zipath(Bites ~ . | 1 , data = DDA_zip, family = "poisson", nlambda=100, lambda.zero.min.ratio=0.005,standardize=T)
As soon as I try to fit a subsample (no replacement) on a new data.frame created with lines selected with sample() from the original df, like shown below, I get the following error
dd.t<-sort(sample(1:nrow(DDA_zip),size=nrow(DDA_zip)*0.8)) DDA_train<-DDA_zip[dd.t,]
fit.r <- zipath(Bites ~ . | 1 , data = DDA_train, family = "poisson", nlambda=100, lambda.zero.min.ratio=0.005,standardize=T)
Errore in zipath_fit(X, Z, Y, weights = weights, offsetx = offsetx, offsetz = offsetz, : NA/NaN/Inf in chiamata a funzione esterna (arg 1)
A similar error is also produced by the cv.zipath() function. Thus I'm not able to CV to find a safer lambda.
Unfortunately, I'm not able to reproduce the error with the bioChemist data set.
I really cannot get the point on this error. Bug? Something related to the data set?
Have you an idea?
Best regards
Giorgio