I apologize if this is an inappropriate question. I thought of it in class the other day, and I couldn't find a specific answer in my textbooks.
I am familiar with the two basic techniques for resampling data:
Subsampling - drawing m observations from a dataset of size n without replacement to generate new samples (where m < n)
Bootstrap - drawing m observations from a dataset of size n with replacement to generate new samples (where m = n)
My question:
Is there a resampling method that blends subsampling with bootstrap? For example, resampling with replacement of every xth observation.
If there isn't, is there a reason why this doesn't work? Would it reduce the advantages of either method without providing any real benefit? I thought it might strike a balance between the bias caused by non-distinct observations in bootstrap samples and the lower power of subsampling. Is this what the .632 Rule is doing?