0

I understood bootstrapping in the statistical context.

Example: we have a sample of 1000 people. We want to know their mean. We pick 5 people at random (with replacement) for 20 times and we compute the mean of the 5 people for each extraction. We end up with 20 means and we can have an estimate of the real mean and of its error. Is this correct?

I was following an online course of machine learning and neural networks. The professor briefly mentioned bootstrapping. How would you use it in the context of machine learning?

I know how bagging is used (ensembles), but I am not sure how plain bootstrapping would be used.

user1315621
  • 133
  • 1
    The duplicate answers your first question, "is this correct?". The short answer is no. – whuber Jul 06 '22 at 16:08
  • I read the duplicated question, but it's not clear to me why the example above would not be correct. Also, I am not sure why my question was labeled as duplicated since I am trying to understand how bootstrapping is used in machine learning (not in plain statistics). Of course, if I didn't understand it in the statistical context, I'd be happy to understand why :) Thank you! – user1315621 Jul 06 '22 at 17:02
  • 1
    You describe a form of resampling, but not bootstrapping. The nature of your distinction between stats and ML is obscure: bootstrapping is the same procedure no matter what you call yourself! – whuber Jul 06 '22 at 17:06
  • Ok, got it. But is my resampling at least a particular example of bootstrapping? Oh, so here's my confusion: when speaking about statistics we usually want to estimate a parameter. How about ML? Why would we use bootstrapping? To estimate the error of the model? If yes, in practice, how? – user1315621 Jul 06 '22 at 17:12
  • Your resampling does not appear to be an example. It's fundamental that the bootstrap resample is obtained from the original sample with replacement and is the same size. Please see some of our posts on bootstrapping. Your characterization of statistics is way, way, too narrow. Exploring some of our more popular posts will help give you some sense of just a few procedures and concepts of statistics. – whuber Jul 06 '22 at 17:15

0 Answers0