[question] how do you control or limit the sample size? about bootstrap.jl HOT 3 CLOSED

All algorithms implemented in the package preserve the size of the input data set. This means that if your input data set consists of N data points, all resampled data sets will contain exactly N data points.
There are bootstrapping algorithms for N out of M resampling but these are not covered by the package.

from bootstrap.jl.

I see! Thank you for the explanation!

How does this work for models that aren't order dependent? Say we're sampling from some set of data points:
[(1, 0.2), (2, 4.5), (3, 6.5), (5, 19)]

and we resample with replacement to

[(5, 19), (1, 0.2), (5, 19), (2, 4.5)]

Is this not the same as effectively removing the (3, 6.5) data point in your sample, and therefore boiling down to jackknifing?

Further, in the previous case, wouldn't many sample sets be equivalent? e.g., [(5, 19), (1, 0.2), (5, 19), (2, 4.5)] = [(5, 19), (5, 19), (1, 0.2), (2, 4.5)]

from bootstrap.jl.

How does this work for models that aren't order dependent? Say we're sampling from some set of data points:
[(1, 0.2), (2, 4.5), (3, 6.5), (5, 19)]
and we resample with replacement to
[(5, 19), (1, 0.2), (5, 19), (2, 4.5)]
Is this not the same as effectively removing the (3, 6.5) data point in your sample, and therefore boiling down to jackknifing?

No. Jack-knifing here gives you

[(1, 0.2), (2, 4.5), (5, 19)]

while bootstrapping gives you e.g.

[(5, 19), (1, 0.2), (5, 19), (2, 4.5)]

(5, 19) are considered to be 2 independent observations here. The moments and even the number of observations of both samples are different.

Further, in the previous case, wouldn't many sample sets be equivalent? e.g., [(5, 19), (1, 0.2), (5, 19), (2, 4.5)] = [(5, 19), (5, 19), (1, 0.2), (2, 4.5)]

They are equivalent in the sense that a statistic that doesn't care about the order of the observations gives you the same result for both samples.
They are different samples for the bootstrap and fundamental to what makes the bootstrap work.

from bootstrap.jl.