post/2021/05/02/the-rms-validate-function/

[utterances-bot]

The {rms} validate function | Random effect

https://randomeffect.net/post/2021/05/02/the-rms-validate-function/

[lijiaqi-github]

Hi, professor Haman, I am learning how to use rms function validate , and found your this post is very useful. If possible, could you explain the meaning of row name in validate , such as index.orig, training test, optimism, index.corrected? I've been searching for relevant information, but haven't found anything. Or could you provide me with some related resources? Thank you.

[mathewrm]

Hello @lijiaqi-github, you can find what the columns mean in Chapter 21.5, p.529. It states that:

Here "training” refers to accuracy when evaluated on the bootstrap sample used to fit the model, and “test” refers to the accuracy when this model is applied without modification to the original sample."

[lijiaqi-github]

Thank you for your reply.

I've read the relevant content but still don't understand it.

The first column shows the original. How to distinguish training and test samples (for example, 70% training and 30% testing)？ What does test mean "without modification to the original sample".

How to understand optimization here. index.corrected seems to be a corrected index.original. The main thing should be to look at index.corrected, right?

I found that index.original - optimism =index.corrected in many parameters, but not all. I'm not sure of the pattern here. I have read the rms book, but maybe because I missed something, I didn't understand many of them well.