Giters

dask / dask-glm

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

First Release Discussion

cicdw opened this issue · comments

commented

@TomAugspurger has mentioned having a first release soon; I'd like to collect the things that we think are blocking that from happening at this moment and check them off the list (any of these left un-checked are open for anyone to work on):

  • some minimal statistical output (e.g., p-values for non-regularized problems)
  • docs
  • some sort of algorithm selection criteria (nothing complicated or fancy, just, e.g., if l1 regularization, don't do newton)
  • add initialization kwarg to all algorithms

Anything else that I'm missing that we should have before a release?

cc: @hussainsultan

commented

I see the docs checked off but there is no link to generated docs in the readme (readthedocs?). I would also add in just a little more info to the readme.

commented

Thanks. Added links in aae15bd

commented

@moody-marlin I think the 4th item (**kwargs to all algos) can be checked off.

Will you have a chance to work on / sketch out what's needed for the statistical output sometime this week? I have some time on Friday that I could use to cut a release.

commented

Hey @TomAugspurger - really sorry about the radio silence on my end; I have time to knock off the other checkboxes before Friday (tonight or tomorrow night). (For the initialization kwarg I just need to tell the algos how to use it).

For the statistical output, I have time to sketch something out for sure, but I don't want to promise a strongly validated set of statistical outputs before Friday, if that makes sense. I.e., it will be a first pass to include a statistic or two, and provide the framework for including more, but it will still need to be checked and double checked to be airtight.

commented

Sounds good. My free time has evaporated (funny how that happens) so it'll be next week at the earliest anyway.

commented

Cool cool, that works for me too (traveling yet again); working on it over here for anyone interested.

commented

late to the party here. I am happy to pick-up some work for the release if it doesnt come in the way of @TomAugspurger and @moody-marlin

commented
commented

I put a 0.1.0 up on PyPI at https://pypi.org/project/dask-glm/

Let me know if you'd like to be added as a maintainer.

commented

@TomAugspurger yea you can put me down as a maintainer!

commented