Skip to content
Prev 521 / 523 Next

[RsR] Welcome to the "R-SIG-Robust" mailing list (Digest mode)

Dear R community,
We analyzed a series of biological samples and obtained results for 60 primary metabolites. For some metabolites, the results are in absolute concentration units, and for others, in relative units, because they were measured using different machines.

The goal is to build a predictive model that selects a small number of metabolites, which can then be measured on one machine to predict a rate associated with each sample. Each sample has a different regeneration rate.

In total, 13 samples were analyzed, each with biological and technical replicates.

First, I performed a forward stepwise regression. Then, I was advised to perform a Leave-One-Out Cross-Validation (LOOCV). I have done this in R, but I am not sure if it is correct, because the resulting predictive model contains only one metabolite.

I would like guidance from the R community on whether my approach and LOOCV implementation are appropriate given this small dataset and the type of data I have. I can provide the R documents and scripts I used so that members can review them and advise whether the analysis is correct.

Any suggestions or recommendations would be greatly appreciated.

Thank you very much for your time and help.

Best regards,
Teresa
Message-ID: <AM0PR03MB6164441660BFED3FE9D06F3CBBEBA@AM0PR03MB6164.eurprd03.prod.outlook.com>
In-Reply-To: <mailman.516.0.1760440594.50025.r-sig-robust@r-project.org>