Precision targets for bootstrap functions

[HDembinski]

I worked out how to iterate the permutation tests until a precision target is reached.

It would be great to implement such a functionality also for the functions

• bootstrap.bias
• bootstrap.bias_corrected
• bootstrap.variance
• bootstrap.confidence_interval

This means adding keywords precision and max_size to the functions and to deprecate size (which would act like max_size=size, precision=0). We cannot do this for bootstrap.bootstrap, because we don't know what the user is computing.

With the keyword return_error (default False) we can optionally return the calculated uncertainty in a backward compatible way.

[amanmdesai]

Hi @HDembinski ,

I would be happy to work on this. Could you please guide me what do I need to do (and also please let me know any references on this topic ) ?

Thanks,
Aman

[HDembinski]

Hi Aman, thank you for your interest and the offer! I am trying to recover my thoughts on this issue, because I eventually abandoned this idea.

The problem is: the number of samples that need to be generated to achieve a given precision grows quadratically e.g. like 1/precision^2. This means you get a huge increase in samples if you go from a precision of 1 % to 0.1 %. I concluded that such an interface would just mislead users to pick a very small precision and then be surprised that it takes a very long time to compute the bootstrap result.

I will close this issue now.

[amanmdesai]

Oh I see.

[amanmdesai]

Hi @HDembinski

Is there anything else that I can contribute to for this package

[HDembinski]

@amanmdesai Apart from the other three open issues I have nothing on my mind right now. All three are not easy targets for a simple contribution, though. If you want to tackle one of them anyway, let me know, I can tell you what should be done.

Otherwise, you may have some ideas on your own what could be improved in resample. Open a new issue if you want to discuss some idea for a new feature or improvement.

If you are not set on contributing to resample, but also open to contributing to other libraries: I am also maintaining iminuit, pyhepmc, numba-stats, and jacobi.

[amanmdesai]

Hi @HDembinski
Thanks for your message.

I have found two issues, one in iminuit and one numba-stats that I would be like to contribute to.