search

malucalle / coda4microbiome

8 stars 0 forks source link

Warning: the standard deviation is zeroWarning: no non-missing arguments to max; returning -Inf$taxa.num #1

Open agatagawad opened 1 year ago

agatagawad commented 1 year ago

Hello,

I have several numeric variables for which I would like to find an explanatory logratio and a microbial signature. But the outcome I get is this one: "Warning: the standard deviation is zeroWarning: no non-missing arguments to max; returning -Inf$taxa.num" . image image

The variable values are like this: image No NAs, nothing below zero. I use the version 0.1.3. What is going there?

I appreciate your help with solving the issue.

Best regards, Agata Gawad

malucalle commented 1 year ago

Hi agatagawad, Your results are probably due to the fact that the method do not find any interesting variable in the variable selection process. You can try two options (separatelly or together): (1) consider a smaller number of folds for the cross-validation (i.e nfolds=4 instead of the default nfolds=10) and (2) lambda = "lambda.min" instead of the default lambda = "lambda.1se"

agatagawad commented 1 year ago

Dear Prof. Calle,

Thank you for your kind reply. Now I understand it. I already realized that those arguments have impact on the outcome. I have two questions:

  1. Can I use coda_glmnet_longitudinal with the continuous outcome? If definitely not, could you give me an alternative?
  2. The argument nvar is not clear to me. What does it change?

Thank you in advance.

Best regards, Agata Gawad

From: Malu Calle @.> Sent: Wednesday, 8 February 2023 18:09 To: malucalle/coda4microbiome @.> Cc: Agata Gawad @.>; Author @.> Subject: Re: [malucalle/coda4microbiome] Warning: the standard deviation is zeroWarning: no non-missing arguments to max; returning -Inf$taxa.num (Issue #1)

Hi agatagawad, Your results are probably due to the fact that the method do not find any interesting variable in the variable selection process. You can try two options (separatelly or together): (1) consider a smaller number of folds for the cross-validation (i.e nfolds=4 instead of the default nfolds=10) and (2) lambda = "lambda.min" instead of the default lambda = "lambda.1se"

— Reply to this email directly, view it on GitHubhttps://github.com/malucalle/coda4microbiome/issues/1#issuecomment-1422959317, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AKMUOLN3ZZGGMNFVQ4PITN3WWPHLZANCNFSM6AAAAAAUO7ON6U. You are receiving this because you authored the thread.Message ID: @.**@.>>