8000 model persistence documentation · Issue #1332 · scikit-learn/scikit-learn · GitHub
[go: up one dir, main page]

Skip to content

model persistence documentation #1332

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
mblondel opened this issue Nov 5, 2012 · 19 comments · Fixed by #3317
Closed

model persistence documentation #1332

mblondel opened this issue Nov 5, 2012 · 19 comments · Fixed by #3317
Labels
Documentation Easy Well-defined and straightforward way to resolve
Milestone

Comments

@mblondel
Copy link
Member
mblondel commented Nov 5, 2012

How to save a model (i.e., model persistence) used to be a sub-section of the tutorial and visible directly from the homepage. But since now it's become a sub-sub-section, it's not visible anymore.

Since it's a very common need it would be nice to make it more prominent. Maybe a section of its own?

@zaxtax
Copy link
Contributor
zaxtax commented Nov 7, 2012

+1

@ogrisel
Copy link
Member
ogrisel commented Nov 7, 2012

+1 too

@AWinterman
Copy link
Contributor

+1 :)

@vene
Copy link
Member
vene commented Jan 6, 2013

Moving just model persistence at the highest level would be a bit out of place. Moving it as a child of a top-level section would be OK, but no current section would accomodate it nicely.

The most elegant thing in my opinion would be move /tutorials/basic_tutorial to /quick_start and saving the tutorials section for in-depth, multi-page ones. Any better ideas?

@amueller
Copy link
Member
amueller commented Jan 6, 2013

Having a "Quick start" section sounds good. The two things under "tutorials" are really quite different in nature.
The old tutorial as quick start could be a good thing to reference for lazy people asking questions on the ml ;)
I can totally understand that someone getting started doesn't want to go through all of the statistical learning tutorial.

@vene
Copy link
Member
vene commented Jan 6, 2013

At the moment, Model Persistence is just a paragraph and that's a bit small for a top-level module. Alternatively, instead of my "Quick Start" suggestion, we could indeed push it up but rename it to something like "Deploying a trained model" and explore it in more detail, like also pointing out how this could sometimes be done avoiding the scikit-learn dependency (I remember this question popping up too)? And also show some real comparison of pickling with cPickle vs joblib (timing, disk usage)?

@jaquesgrobler
Copy link
Member

So what's the verdict on this issue?
One thing - moving the basic tutorial out of the Tutorials section and into a Quick start section sounds good and would definitely be quick and easy, but won't the other tutorial look a tad lonely then? (Not that that is a major issue).. @vene 's second suggestion also sounds good to me

@vene
Copy link
Member
vene commented Jan 11, 2013

Wasn't there another tutorial that we were planning to merge? This would
make the big one less lonely :)

On Fri, Jan 11, 2013 at 2:11 PM, Jaques Grobler notifications@github.comwrote:

So what's the verdict on this issue?
One thing - moving the basic tutorial out of the Tutorials section and
into a Quick start section sounds good and would definitely be quick and
easy, but won't the other tutorial look a tad lonely then? (Not that that
is a major issue).. @vene https://github.com/vene 's second suggestion
also sounds good to me


Reply to this email directly or view it on GitHubhttps://github.com//issues/1332#issuecomment-12145498.

@jaquesgrobler
Copy link
Member

Hmm good question.. perhaps they were the external tutorials (the astro and neuroimg ones). Can't think of any other ones.. @amueller @GaelVaroquaux You guys maybe know?

@amueller
Copy link
Member

There is @ogrisel's on text processing but I think that is pretty out of date.

@ogrisel
Copy link
Member
ogrisel commented Jan 12, 2013

The narrative doc on text feature expression is now good enough even though it does not feature how to pipeline it with a classifier (but there is an example).

@amueller
Copy link
Member

OT @ogrisel the tutorial is still online and uses an old api, right? maybe you should take it down. I got questions about it on IRC recently.

@amueller
Copy link
Member

Reassigned to next release as I don't really see a consent.

@ogrisel
Copy link
Member
ogrisel commented Jan 21, 2013

I will open a new issue to reintegrate / update the text data processing part of the tutorial into a new tutorial section in the scikit-learn narrative documentation. The rest of the tutorial can be dropped as redundant.

@amueller
Copy link
Member

Thanks :)

@amueller
Copy link
Member

retagging 0.14

@GaelVaroquaux
Copy link
Member

No progress. Retagging.

@kaushik94
Copy link
Contributor

i guess i can do a PR for this.
shall i go with the "quick start" model or any other ideas are to be considered ?

@jnothman
Copy link
Member

Fixed by #3317

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Documentation Easy Well-defined and straightforward way to resolve
Projects
None yet
Development

Successfully merging a pull request may close this issue.

10 participants
0