Accuracy, sensitivity, and specificity

Dear MicOmixs team,

I want to evaluate the performance of my sPLD-DA based on differents VIP values. For this, I would like to know how to obtain the accuracy, sensitivity, and specificity values. As in this paper, https://static-content.springer.com/esm/art%3A10.1038%2Fs41598-018-21763-6/MediaObjects/41598_2018_21763_MOESM1_ESM.pdf, where the authors used MixOmics and chechek it.

Thanks very much in advance.
All the help is much appreciated.
Regards,
Patricia

Hi Patricia,

Thanks for using mixOmics!

I’m not sure about the details of how the mentioned paper extracted the values, but we calculate these values internally in statauc internal function as you can see: https://github.com/mixOmicsTeam/mixOmics/blob/f183656a825e83e7f3b16092b21c2f13c5799b9d/R/roc_utils.R#L27.
Basically, auroc.mixo_splsda : https://github.com/mixOmicsTeam/mixOmics/blob/f183656a825e83e7f3b16092b21c2f13c5799b9d/R/AUC_ROC.R#L3
calls this function to evaluate the AUC of the model. You can customise statauc function to extract the df as well, which will give you the Sensitivity and Specificity of the model. Accuracy of the model will also be, by definition, 1- error_rate. The perf function provides the model error rate.

Hope it helps

Al

Hi AI,

Thanks very much for your reply. It worked!

Cheers,
Patricia