Usage in practised application about forest-confidence-interval

[arokem]

Sorry - I don't understand the question. What do you mean by "how we support to use the error bar"? You can plot it, as we show in the examples, but the number itself can be used in any type of inference that you usually do with error estimates.

[JIAZHEN]

Hi @arokem, Sorry, type - support -> suppose.

• With the number, how would we measure the confidence interval of Random Forest?
• In the example, the number named Standard deviation, what is this Standard deviation to?
• do you have an example to say: here is my predict probability for a X, model predicted it as class Foo, by using this number I can say I'm 95%~100% confident the prediction is correct?

Thanks for answering

[arokem]

I'd recommend reading the Wager paper on which this is based, and which explains all this: http://jmlr.org/papers/volume15/wager14a/wager14a.pdf

…

On Fri, Dec 22, 2017 at 1:45 AM, Jiazhen Xie ***@***.***> wrote: Hi @arokem <https://github.com/arokem>, Sorry, type - support -> suppose. - With the number, how would we measure the confidence interval of Random Forest? - In the example, the number named Standard deviation, what is this Standard deviation to? - do you have an example to say: here is my predict probability for a X, model predicted it as class Foo, by using this number I can say I'm 95%~100% confident the prediction is correct? Thanks for answering — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#71 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAHPNjebNc2oJkfTM6hWpT9Tg8xFS6Xpks5tC3ohgaJpZM4RIrXW> .

[JIAZHEN]

Though there could be existing examples to demonstrate the paper, I guess you're right to read it. Thanks for the replies.