[SciPy-Dev] Adding non-parametric methods to scipy.stats

[Hans Dembinski]

Dear all,

> On 11. Jun 2020, at 07:46, Romain Jacob <jacobr at ethz.ch> wrote:
> 
> I think a dedicated function makes more sense. This function takes as input an array, a percentile and a confidence level, and returns the corresponding one-sided confidence intervals.
> 
> I quickly looked at the list of existing functions in scipy.stats but did not see any function in "summary statistics" that does similar things. So I would go for a new function.

I just joined the list, so I apologise for any etiquette-breaking in advance, but would like  to inject here that I am collaborating with Daniel Saxton on `resample`, a library that implements the jackknife and bootstrap, which can be used - among many other things - to compute confidence intervals for quantiles/percentiles.

https://github.com/dsaxton/resample

We are currently working on interface and documentation and adding more unit tests and benchmarks, but `resample` is already the most complete library that implements resampling methods in Python. 

Seeing that https://github.com/scipy/scipy/issues/10577 explicitly mentions bootstrapping, we are interested in merging our work into scipy. We use the BSD 3-clause license, so the license should not be an issue. Is there already work ongoing on bootstrap methods? With whom should we collaborate?

Some context about us:

Daniel is a data analyst working in the financial industry. I am a particle physicist and the author of Boost Histogram (C++ and Python, https://github.com/boostorg/histogram, https://github.com/scikit-hep/boost-histogram) and the maintainer of iminuit, the general purpose minimiser and error computer (C++ and Python, https://github.com/scikit-hep/iminuit).

Best regards,
Hans