You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@mahout.apache.org by Andrew Palumbo <ap...@outlook.com> on 2014/10/24 00:57:39 UTC
Mahout 1.0 features (revisited)
I made a list of Mahout 1.0 features by engine a while back which I’ve just reorganized (and added links to). I’m not sure whether It’s more confusing this way or the original way:
Original:
https://andrewpalumbo.github.io/algorithms_by_engine
One concern is that it doesn’t emphasize the new work done in math-scala enough- given that there’s only one line of the table devoted to it. For this reason, I’ve reorganized it a bit.
Reorganized:
https://andrewpalumbo.github.io/features_by_engine
But this becomes confusing- eg. having 2 entries for SSVD- one in “Mahout Math-Scala Core and Scala DSL” and one in “Collaborative Filtering with CLI Drivers”. For this reason I’ve tried to break it down into core-algorithms (math-scala, dsl, etc), and and end2end implementations of those algorithms (and others) with a CLI. Here though too, I feel that this is somewhat confusing: eg. does it seem that there is no SSVD algorithm implemented for Spark or H2O?
I’d like to commit this in place of the “List of algorithms page”:
http://mahout.apache.org/users/basics/algorithms.html
This new table is inclusive of all that is on the list for 0.9 (with the exception of FPG).
Briefly going over yesterday’s “Mahout vs Spark” thread: http://mail-archives.us.apache.org/mod_mbox/mahout-user/201410.mbox/browser,I’d like to put something out there to give users, developers and contributors a quick overview of what Mahout 1.0 is/has/does/extends/keeps/can be used for. However I don’t want it to cause any more confusion.
There are probably some poorly described features, and maybe some incorrect links, so once I’ve committed it, you guys may want to have at it.
Or I can just commit as is and people can have at the organization.
I’m in favor of the revised/reorganized page. I’d appreciate feedback from any and all- specifically on whether the page is confusing and if so how it might be better organized.
Thanks,
Andy
Re: Mahout 1.0 features (revisited)
Posted by Dmitriy Lyubimov <dl...@gmail.com>.
Ah, you've already added links behind the X marks .. :) ok awesome.
On Fri, Oct 24, 2014 at 11:23 AM, Dmitriy Lyubimov <dl...@gmail.com>
wrote:
> This is so awesome.
>
> i think i need to re(move) issues links from scala spark bindings page and
> just point to this page. Perhaps it is good to add the links to the jiras,
> at least for WIP entries
>
> On Fri, Oct 24, 2014 at 11:19 AM, Andrew Palumbo <ap...@outlook.com>
> wrote:
>
>> aha.. yes.. Fixed it.. thx. and please do feel free to hack away at any
>> content/organization that you see fit. I've tried to give a good
>> description of Math-Scala/DSL all in one section.
>>
>> thx
>>
>> > Date: Fri, 24 Oct 2014 11:01:00 -0700
>> > Subject: Re: Mahout 1.0 features (revisited)
>> > From: dlieu.7@gmail.com
>> > To: dev@mahout.apache.org
>> >
>> > awesome
>> >
>> > (typo in "Collaborative filtering" (have i spelled it right?) :)
>> >
>> >
>> > On Fri, Oct 24, 2014 at 10:17 AM, Andrew Palumbo <ap...@outlook.com>
>> wrote:
>> >
>> > > ok, committed here:
>> > >
>> > > http://mahout.apache.org/users/basics/algorithms.html
>> > >
>> > > thx
>> > >
>> > > > From: ap.dev@outlook.com
>> > > > To: dev@mahout.apache.org
>> > > > Subject: RE: Mahout 1.0 features (revisited)
>> > > > Date: Fri, 24 Oct 2014 11:49:15 -0400
>> > > >
>> > > > Thank you guys for the feedback! I'll just commit the reorganized
>> page
>> > > in place of the current "List of Algorithms" page as is and the
>> experts on
>> > > each section can have at it / reorganize as they see fit.
>> > > >
>> > > > I know that this page is not a big deal but my goal is to clarify-
>> not
>> > > confuse- so I really appreciate the eyes!
>> > > >
>> > > > Another concern is that the page did not reflect the Hierarchical
>> design
>> > > of the new code.. ie. the Engine-bindings inheriting from and
>> optimizing
>> > > math-scala, which I think is a huge Mahout strong point. Maybe that
>> is for
>> > > another page...
>> > > >
>> > > > Thanks for the comments, Thejas...
>> > > > > > > I guess another thing that could be added would be (needs
>> > > > > > > development) so that developer can dig in and start working.
>> > > >
>> > > > this is a good idea but might clutter the page, Instead maybe we
>> can
>> > > make it known that anything (non- MR) without a marker "needs
>> > > development". Then maybe an explanation/link from the "How to
>> Contribute
>> > > page" regarding this fact.
>> > > >
>> > > > > > > I think this is a very clear, and another thing we can add is
>> maybe
>> > > > > > > a link to the source code of the algorithm or an
>> example/tutorial.
>> > > >
>> > > > I like this idea very much- maybe moving the doc links over to the
>> algo
>> > > column and then having the markers link to directly to the github
>> source.
>> > > The only problem i can think of here is that eg. "Item-Based
>> Collaborative
>> > > Filtering" has docs for both MapReduce and Spark, and in the future
>> this
>> > > may occur more often- though with the legacy status of MapReduce and
>> the
>> > > recent abstraction of CLI drivers, probably won't.
>> > > >
>> > > > It would be nice to have another dimension to work in.
>> > > >
>> > > > Thanks again,
>> > > >
>> > > > Andy
>> > > >
>> > > > From: ap.dev@outlook.com
>> > > > To: dev@mahout.apache.org
>> > > > Subject: RE: Mahout 1.0 features (revisited)
>> > > > Date: Fri, 24 Oct 2014 11:46:34 -0400
>> > > >
>> > > >
>> > > >
>> > > >
>> > > >
>> > > >
>> > > >
>> > > > Thank you guys for the feedback! I'll just commit the reorganized
>> page
>> > > in place of the current "List of Algorithms" page as is and the
>> experts on
>> > > each section can have at it / reorganize as they see fit.
>> > > >
>> > > > I know that this page is not a big deal but my goal is to clarify-
>> not
>> > > confuse- so I really appreciate the eyes!
>> > > >
>> > > > Another concern is that the page did not reflect the Hierarchical
>> design
>> > > of the new code.. ie. the Engine-bindings inheriting from and
>> optimizing
>> > > math-scala, which I think is a huge Mahout strong point. Maybe that
>> is for
>> > > another page...
>> > > >
>> > > > Thanks for the comments, Thejas...
>> > > > > > > I guess another thing that could be added would be (needs
>> > > > > > > development) so that developer can dig in and start
>> working.this
>> > > is a good idea but might clutter the page, Instead maybe we can make
>> it
>> > > known that anything (non- MR) without a marker "needs development".
>> Then
>> > > maybe an explanation/link from the "How to Contribute page" regarding
>> this
>> > > fact.
>> > > >
>> > > > > > > I think this is a very clear, and another thing we can add is
>> maybe
>> > > > > > > a link to the source code of the algorithm or an
>> > > example/tutorial.I like this idea very much- maybe moving the doc
>> links
>> > > over to the algo column and then having the markers link to directly
>> to the
>> > > github source. The only problem i can think of here is that eg.
>> > > "Item-Based Collaborative Filtering" has docs for both MapReduce and
>> Spark,
>> > > and in the future this may occur more often- though with the legacy
>> status
>> > > of MapReduce and the recent abstraction of CLI drivers, probably
>> won't.
>> > > >
>> > > > It would be nice to have another dimension to work in.
>> > > >
>> > > > Thanks again,
>> > > >
>> > > > Andy
>> > > >
>> > > >
>> > > > > Date: Fri, 24 Oct 2014 10:54:09 +0200
>> > > > > Subject: Re: Mahout 1.0 features (revisited)
>> > > > > From: frank@frankscholten.nl
>> > > > > To: dev@mahout.apache.org
>> > > > >
>> > > > > Great!
>> > > > >
>> > > > > On Fri, Oct 24, 2014 at 8:01 AM, Ted Dunning <
>> ted.dunning@gmail.com>
>> > > wrote:
>> > > > >
>> > > > > > On Thu, Oct 23, 2014 at 3:57 PM, Andrew Palumbo <
>> ap.dev@outlook.com>
>> > > > > > wrote:
>> > > > > >
>> > > > > > > Or I can just commit as is and people can have at the
>> organization.
>> > > > > > >
>> > > > > >
>> > > > > >
>> > > > > > Sounds good to me!
>> > > > > >
>> > > >
>> > > >
>> > >
>> > >
>>
>>
>
>
Re: Mahout 1.0 features (revisited)
Posted by Dmitriy Lyubimov <dl...@gmail.com>.
This is so awesome.
i think i need to re(move) issues links from scala spark bindings page and
just point to this page. Perhaps it is good to add the links to the jiras,
at least for WIP entries
On Fri, Oct 24, 2014 at 11:19 AM, Andrew Palumbo <ap...@outlook.com> wrote:
> aha.. yes.. Fixed it.. thx. and please do feel free to hack away at any
> content/organization that you see fit. I've tried to give a good
> description of Math-Scala/DSL all in one section.
>
> thx
>
> > Date: Fri, 24 Oct 2014 11:01:00 -0700
> > Subject: Re: Mahout 1.0 features (revisited)
> > From: dlieu.7@gmail.com
> > To: dev@mahout.apache.org
> >
> > awesome
> >
> > (typo in "Collaborative filtering" (have i spelled it right?) :)
> >
> >
> > On Fri, Oct 24, 2014 at 10:17 AM, Andrew Palumbo <ap...@outlook.com>
> wrote:
> >
> > > ok, committed here:
> > >
> > > http://mahout.apache.org/users/basics/algorithms.html
> > >
> > > thx
> > >
> > > > From: ap.dev@outlook.com
> > > > To: dev@mahout.apache.org
> > > > Subject: RE: Mahout 1.0 features (revisited)
> > > > Date: Fri, 24 Oct 2014 11:49:15 -0400
> > > >
> > > > Thank you guys for the feedback! I'll just commit the reorganized
> page
> > > in place of the current "List of Algorithms" page as is and the
> experts on
> > > each section can have at it / reorganize as they see fit.
> > > >
> > > > I know that this page is not a big deal but my goal is to clarify-
> not
> > > confuse- so I really appreciate the eyes!
> > > >
> > > > Another concern is that the page did not reflect the Hierarchical
> design
> > > of the new code.. ie. the Engine-bindings inheriting from and
> optimizing
> > > math-scala, which I think is a huge Mahout strong point. Maybe that
> is for
> > > another page...
> > > >
> > > > Thanks for the comments, Thejas...
> > > > > > > I guess another thing that could be added would be (needs
> > > > > > > development) so that developer can dig in and start working.
> > > >
> > > > this is a good idea but might clutter the page, Instead maybe we can
> > > make it known that anything (non- MR) without a marker "needs
> > > development". Then maybe an explanation/link from the "How to
> Contribute
> > > page" regarding this fact.
> > > >
> > > > > > > I think this is a very clear, and another thing we can add is
> maybe
> > > > > > > a link to the source code of the algorithm or an
> example/tutorial.
> > > >
> > > > I like this idea very much- maybe moving the doc links over to the
> algo
> > > column and then having the markers link to directly to the github
> source.
> > > The only problem i can think of here is that eg. "Item-Based
> Collaborative
> > > Filtering" has docs for both MapReduce and Spark, and in the future
> this
> > > may occur more often- though with the legacy status of MapReduce and
> the
> > > recent abstraction of CLI drivers, probably won't.
> > > >
> > > > It would be nice to have another dimension to work in.
> > > >
> > > > Thanks again,
> > > >
> > > > Andy
> > > >
> > > > From: ap.dev@outlook.com
> > > > To: dev@mahout.apache.org
> > > > Subject: RE: Mahout 1.0 features (revisited)
> > > > Date: Fri, 24 Oct 2014 11:46:34 -0400
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > > Thank you guys for the feedback! I'll just commit the reorganized
> page
> > > in place of the current "List of Algorithms" page as is and the
> experts on
> > > each section can have at it / reorganize as they see fit.
> > > >
> > > > I know that this page is not a big deal but my goal is to clarify-
> not
> > > confuse- so I really appreciate the eyes!
> > > >
> > > > Another concern is that the page did not reflect the Hierarchical
> design
> > > of the new code.. ie. the Engine-bindings inheriting from and
> optimizing
> > > math-scala, which I think is a huge Mahout strong point. Maybe that
> is for
> > > another page...
> > > >
> > > > Thanks for the comments, Thejas...
> > > > > > > I guess another thing that could be added would be (needs
> > > > > > > development) so that developer can dig in and start
> working.this
> > > is a good idea but might clutter the page, Instead maybe we can make
> it
> > > known that anything (non- MR) without a marker "needs development".
> Then
> > > maybe an explanation/link from the "How to Contribute page" regarding
> this
> > > fact.
> > > >
> > > > > > > I think this is a very clear, and another thing we can add is
> maybe
> > > > > > > a link to the source code of the algorithm or an
> > > example/tutorial.I like this idea very much- maybe moving the doc links
> > > over to the algo column and then having the markers link to directly
> to the
> > > github source. The only problem i can think of here is that eg.
> > > "Item-Based Collaborative Filtering" has docs for both MapReduce and
> Spark,
> > > and in the future this may occur more often- though with the legacy
> status
> > > of MapReduce and the recent abstraction of CLI drivers, probably won't.
> > > >
> > > > It would be nice to have another dimension to work in.
> > > >
> > > > Thanks again,
> > > >
> > > > Andy
> > > >
> > > >
> > > > > Date: Fri, 24 Oct 2014 10:54:09 +0200
> > > > > Subject: Re: Mahout 1.0 features (revisited)
> > > > > From: frank@frankscholten.nl
> > > > > To: dev@mahout.apache.org
> > > > >
> > > > > Great!
> > > > >
> > > > > On Fri, Oct 24, 2014 at 8:01 AM, Ted Dunning <
> ted.dunning@gmail.com>
> > > wrote:
> > > > >
> > > > > > On Thu, Oct 23, 2014 at 3:57 PM, Andrew Palumbo <
> ap.dev@outlook.com>
> > > > > > wrote:
> > > > > >
> > > > > > > Or I can just commit as is and people can have at the
> organization.
> > > > > > >
> > > > > >
> > > > > >
> > > > > > Sounds good to me!
> > > > > >
> > > >
> > > >
> > >
> > >
>
>
RE: Mahout 1.0 features (revisited)
Posted by Andrew Palumbo <ap...@outlook.com>.
aha.. yes.. Fixed it.. thx. and please do feel free to hack away at any content/organization that you see fit. I've tried to give a good description of Math-Scala/DSL all in one section.
thx
> Date: Fri, 24 Oct 2014 11:01:00 -0700
> Subject: Re: Mahout 1.0 features (revisited)
> From: dlieu.7@gmail.com
> To: dev@mahout.apache.org
>
> awesome
>
> (typo in "Collaborative filtering" (have i spelled it right?) :)
>
>
> On Fri, Oct 24, 2014 at 10:17 AM, Andrew Palumbo <ap...@outlook.com> wrote:
>
> > ok, committed here:
> >
> > http://mahout.apache.org/users/basics/algorithms.html
> >
> > thx
> >
> > > From: ap.dev@outlook.com
> > > To: dev@mahout.apache.org
> > > Subject: RE: Mahout 1.0 features (revisited)
> > > Date: Fri, 24 Oct 2014 11:49:15 -0400
> > >
> > > Thank you guys for the feedback! I'll just commit the reorganized page
> > in place of the current "List of Algorithms" page as is and the experts on
> > each section can have at it / reorganize as they see fit.
> > >
> > > I know that this page is not a big deal but my goal is to clarify- not
> > confuse- so I really appreciate the eyes!
> > >
> > > Another concern is that the page did not reflect the Hierarchical design
> > of the new code.. ie. the Engine-bindings inheriting from and optimizing
> > math-scala, which I think is a huge Mahout strong point. Maybe that is for
> > another page...
> > >
> > > Thanks for the comments, Thejas...
> > > > > > I guess another thing that could be added would be (needs
> > > > > > development) so that developer can dig in and start working.
> > >
> > > this is a good idea but might clutter the page, Instead maybe we can
> > make it known that anything (non- MR) without a marker "needs
> > development". Then maybe an explanation/link from the "How to Contribute
> > page" regarding this fact.
> > >
> > > > > > I think this is a very clear, and another thing we can add is maybe
> > > > > > a link to the source code of the algorithm or an example/tutorial.
> > >
> > > I like this idea very much- maybe moving the doc links over to the algo
> > column and then having the markers link to directly to the github source.
> > The only problem i can think of here is that eg. "Item-Based Collaborative
> > Filtering" has docs for both MapReduce and Spark, and in the future this
> > may occur more often- though with the legacy status of MapReduce and the
> > recent abstraction of CLI drivers, probably won't.
> > >
> > > It would be nice to have another dimension to work in.
> > >
> > > Thanks again,
> > >
> > > Andy
> > >
> > > From: ap.dev@outlook.com
> > > To: dev@mahout.apache.org
> > > Subject: RE: Mahout 1.0 features (revisited)
> > > Date: Fri, 24 Oct 2014 11:46:34 -0400
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > > Thank you guys for the feedback! I'll just commit the reorganized page
> > in place of the current "List of Algorithms" page as is and the experts on
> > each section can have at it / reorganize as they see fit.
> > >
> > > I know that this page is not a big deal but my goal is to clarify- not
> > confuse- so I really appreciate the eyes!
> > >
> > > Another concern is that the page did not reflect the Hierarchical design
> > of the new code.. ie. the Engine-bindings inheriting from and optimizing
> > math-scala, which I think is a huge Mahout strong point. Maybe that is for
> > another page...
> > >
> > > Thanks for the comments, Thejas...
> > > > > > I guess another thing that could be added would be (needs
> > > > > > development) so that developer can dig in and start working.this
> > is a good idea but might clutter the page, Instead maybe we can make it
> > known that anything (non- MR) without a marker "needs development". Then
> > maybe an explanation/link from the "How to Contribute page" regarding this
> > fact.
> > >
> > > > > > I think this is a very clear, and another thing we can add is maybe
> > > > > > a link to the source code of the algorithm or an
> > example/tutorial.I like this idea very much- maybe moving the doc links
> > over to the algo column and then having the markers link to directly to the
> > github source. The only problem i can think of here is that eg.
> > "Item-Based Collaborative Filtering" has docs for both MapReduce and Spark,
> > and in the future this may occur more often- though with the legacy status
> > of MapReduce and the recent abstraction of CLI drivers, probably won't.
> > >
> > > It would be nice to have another dimension to work in.
> > >
> > > Thanks again,
> > >
> > > Andy
> > >
> > >
> > > > Date: Fri, 24 Oct 2014 10:54:09 +0200
> > > > Subject: Re: Mahout 1.0 features (revisited)
> > > > From: frank@frankscholten.nl
> > > > To: dev@mahout.apache.org
> > > >
> > > > Great!
> > > >
> > > > On Fri, Oct 24, 2014 at 8:01 AM, Ted Dunning <te...@gmail.com>
> > wrote:
> > > >
> > > > > On Thu, Oct 23, 2014 at 3:57 PM, Andrew Palumbo <ap...@outlook.com>
> > > > > wrote:
> > > > >
> > > > > > Or I can just commit as is and people can have at the organization.
> > > > > >
> > > > >
> > > > >
> > > > > Sounds good to me!
> > > > >
> > >
> > >
> >
> >
Re: Mahout 1.0 features (revisited)
Posted by Dmitriy Lyubimov <dl...@gmail.com>.
awesome
(typo in "Collaborative filtering" (have i spelled it right?) :)
On Fri, Oct 24, 2014 at 10:17 AM, Andrew Palumbo <ap...@outlook.com> wrote:
> ok, committed here:
>
> http://mahout.apache.org/users/basics/algorithms.html
>
> thx
>
> > From: ap.dev@outlook.com
> > To: dev@mahout.apache.org
> > Subject: RE: Mahout 1.0 features (revisited)
> > Date: Fri, 24 Oct 2014 11:49:15 -0400
> >
> > Thank you guys for the feedback! I'll just commit the reorganized page
> in place of the current "List of Algorithms" page as is and the experts on
> each section can have at it / reorganize as they see fit.
> >
> > I know that this page is not a big deal but my goal is to clarify- not
> confuse- so I really appreciate the eyes!
> >
> > Another concern is that the page did not reflect the Hierarchical design
> of the new code.. ie. the Engine-bindings inheriting from and optimizing
> math-scala, which I think is a huge Mahout strong point. Maybe that is for
> another page...
> >
> > Thanks for the comments, Thejas...
> > > > > I guess another thing that could be added would be (needs
> > > > > development) so that developer can dig in and start working.
> >
> > this is a good idea but might clutter the page, Instead maybe we can
> make it known that anything (non- MR) without a marker "needs
> development". Then maybe an explanation/link from the "How to Contribute
> page" regarding this fact.
> >
> > > > > I think this is a very clear, and another thing we can add is maybe
> > > > > a link to the source code of the algorithm or an example/tutorial.
> >
> > I like this idea very much- maybe moving the doc links over to the algo
> column and then having the markers link to directly to the github source.
> The only problem i can think of here is that eg. "Item-Based Collaborative
> Filtering" has docs for both MapReduce and Spark, and in the future this
> may occur more often- though with the legacy status of MapReduce and the
> recent abstraction of CLI drivers, probably won't.
> >
> > It would be nice to have another dimension to work in.
> >
> > Thanks again,
> >
> > Andy
> >
> > From: ap.dev@outlook.com
> > To: dev@mahout.apache.org
> > Subject: RE: Mahout 1.0 features (revisited)
> > Date: Fri, 24 Oct 2014 11:46:34 -0400
> >
> >
> >
> >
> >
> >
> >
> > Thank you guys for the feedback! I'll just commit the reorganized page
> in place of the current "List of Algorithms" page as is and the experts on
> each section can have at it / reorganize as they see fit.
> >
> > I know that this page is not a big deal but my goal is to clarify- not
> confuse- so I really appreciate the eyes!
> >
> > Another concern is that the page did not reflect the Hierarchical design
> of the new code.. ie. the Engine-bindings inheriting from and optimizing
> math-scala, which I think is a huge Mahout strong point. Maybe that is for
> another page...
> >
> > Thanks for the comments, Thejas...
> > > > > I guess another thing that could be added would be (needs
> > > > > development) so that developer can dig in and start working.this
> is a good idea but might clutter the page, Instead maybe we can make it
> known that anything (non- MR) without a marker "needs development". Then
> maybe an explanation/link from the "How to Contribute page" regarding this
> fact.
> >
> > > > > I think this is a very clear, and another thing we can add is maybe
> > > > > a link to the source code of the algorithm or an
> example/tutorial.I like this idea very much- maybe moving the doc links
> over to the algo column and then having the markers link to directly to the
> github source. The only problem i can think of here is that eg.
> "Item-Based Collaborative Filtering" has docs for both MapReduce and Spark,
> and in the future this may occur more often- though with the legacy status
> of MapReduce and the recent abstraction of CLI drivers, probably won't.
> >
> > It would be nice to have another dimension to work in.
> >
> > Thanks again,
> >
> > Andy
> >
> >
> > > Date: Fri, 24 Oct 2014 10:54:09 +0200
> > > Subject: Re: Mahout 1.0 features (revisited)
> > > From: frank@frankscholten.nl
> > > To: dev@mahout.apache.org
> > >
> > > Great!
> > >
> > > On Fri, Oct 24, 2014 at 8:01 AM, Ted Dunning <te...@gmail.com>
> wrote:
> > >
> > > > On Thu, Oct 23, 2014 at 3:57 PM, Andrew Palumbo <ap...@outlook.com>
> > > > wrote:
> > > >
> > > > > Or I can just commit as is and people can have at the organization.
> > > > >
> > > >
> > > >
> > > > Sounds good to me!
> > > >
> >
> >
>
>
RE: Mahout 1.0 features (revisited)
Posted by Andrew Palumbo <ap...@outlook.com>.
ok, committed here:
http://mahout.apache.org/users/basics/algorithms.html
thx
> From: ap.dev@outlook.com
> To: dev@mahout.apache.org
> Subject: RE: Mahout 1.0 features (revisited)
> Date: Fri, 24 Oct 2014 11:49:15 -0400
>
> Thank you guys for the feedback! I'll just commit the reorganized page in place of the current "List of Algorithms" page as is and the experts on each section can have at it / reorganize as they see fit.
>
> I know that this page is not a big deal but my goal is to clarify- not confuse- so I really appreciate the eyes!
>
> Another concern is that the page did not reflect the Hierarchical design of the new code.. ie. the Engine-bindings inheriting from and optimizing math-scala, which I think is a huge Mahout strong point. Maybe that is for another page...
>
> Thanks for the comments, Thejas...
> > > > I guess another thing that could be added would be (needs
> > > > development) so that developer can dig in and start working.
>
> this is a good idea but might clutter the page, Instead maybe we can make it known that anything (non- MR) without a marker "needs development". Then maybe an explanation/link from the "How to Contribute page" regarding this fact.
>
> > > > I think this is a very clear, and another thing we can add is maybe
> > > > a link to the source code of the algorithm or an example/tutorial.
>
> I like this idea very much- maybe moving the doc links over to the algo column and then having the markers link to directly to the github source. The only problem i can think of here is that eg. "Item-Based Collaborative Filtering" has docs for both MapReduce and Spark, and in the future this may occur more often- though with the legacy status of MapReduce and the recent abstraction of CLI drivers, probably won't.
>
> It would be nice to have another dimension to work in.
>
> Thanks again,
>
> Andy
>
> From: ap.dev@outlook.com
> To: dev@mahout.apache.org
> Subject: RE: Mahout 1.0 features (revisited)
> Date: Fri, 24 Oct 2014 11:46:34 -0400
>
>
>
>
>
>
>
> Thank you guys for the feedback! I'll just commit the reorganized page in place of the current "List of Algorithms" page as is and the experts on each section can have at it / reorganize as they see fit.
>
> I know that this page is not a big deal but my goal is to clarify- not confuse- so I really appreciate the eyes!
>
> Another concern is that the page did not reflect the Hierarchical design of the new code.. ie. the Engine-bindings inheriting from and optimizing math-scala, which I think is a huge Mahout strong point. Maybe that is for another page...
>
> Thanks for the comments, Thejas...
> > > > I guess another thing that could be added would be (needs
> > > > development) so that developer can dig in and start working.this is a good idea but might clutter the page, Instead maybe we can make it known that anything (non- MR) without a marker "needs development". Then maybe an explanation/link from the "How to Contribute page" regarding this fact.
>
> > > > I think this is a very clear, and another thing we can add is maybe
> > > > a link to the source code of the algorithm or an example/tutorial.I like this idea very much- maybe moving the doc links over to the algo column and then having the markers link to directly to the github source. The only problem i can think of here is that eg. "Item-Based Collaborative Filtering" has docs for both MapReduce and Spark, and in the future this may occur more often- though with the legacy status of MapReduce and the recent abstraction of CLI drivers, probably won't.
>
> It would be nice to have another dimension to work in.
>
> Thanks again,
>
> Andy
>
>
> > Date: Fri, 24 Oct 2014 10:54:09 +0200
> > Subject: Re: Mahout 1.0 features (revisited)
> > From: frank@frankscholten.nl
> > To: dev@mahout.apache.org
> >
> > Great!
> >
> > On Fri, Oct 24, 2014 at 8:01 AM, Ted Dunning <te...@gmail.com> wrote:
> >
> > > On Thu, Oct 23, 2014 at 3:57 PM, Andrew Palumbo <ap...@outlook.com>
> > > wrote:
> > >
> > > > Or I can just commit as is and people can have at the organization.
> > > >
> > >
> > >
> > > Sounds good to me!
> > >
>
>
RE: Mahout 1.0 features (revisited)
Posted by Andrew Palumbo <ap...@outlook.com>.
Thank you guys for the feedback! I'll just commit the reorganized page in place of the current "List of Algorithms" page as is and the experts on each section can have at it / reorganize as they see fit.
I know that this page is not a big deal but my goal is to clarify- not confuse- so I really appreciate the eyes!
Another concern is that the page did not reflect the Hierarchical design of the new code.. ie. the Engine-bindings inheriting from and optimizing math-scala, which I think is a huge Mahout strong point. Maybe that is for another page...
Thanks for the comments, Thejas...
> > > I guess another thing that could be added would be (needs
> > > development) so that developer can dig in and start working.
this is a good idea but might clutter the page, Instead maybe we can make it known that anything (non- MR) without a marker "needs development". Then maybe an explanation/link from the "How to Contribute page" regarding this fact.
> > > I think this is a very clear, and another thing we can add is maybe
> > > a link to the source code of the algorithm or an example/tutorial.
I like this idea very much- maybe moving the doc links over to the algo column and then having the markers link to directly to the github source. The only problem i can think of here is that eg. "Item-Based Collaborative Filtering" has docs for both MapReduce and Spark, and in the future this may occur more often- though with the legacy status of MapReduce and the recent abstraction of CLI drivers, probably won't.
It would be nice to have another dimension to work in.
Thanks again,
Andy
From: ap.dev@outlook.com
To: dev@mahout.apache.org
Subject: RE: Mahout 1.0 features (revisited)
Date: Fri, 24 Oct 2014 11:46:34 -0400
Thank you guys for the feedback! I'll just commit the reorganized page in place of the current "List of Algorithms" page as is and the experts on each section can have at it / reorganize as they see fit.
I know that this page is not a big deal but my goal is to clarify- not confuse- so I really appreciate the eyes!
Another concern is that the page did not reflect the Hierarchical design of the new code.. ie. the Engine-bindings inheriting from and optimizing math-scala, which I think is a huge Mahout strong point. Maybe that is for another page...
Thanks for the comments, Thejas...
> > > I guess another thing that could be added would be (needs
> > > development) so that developer can dig in and start working.this is a good idea but might clutter the page, Instead maybe we can make it known that anything (non- MR) without a marker "needs development". Then maybe an explanation/link from the "How to Contribute page" regarding this fact.
> > > I think this is a very clear, and another thing we can add is maybe
> > > a link to the source code of the algorithm or an example/tutorial.I like this idea very much- maybe moving the doc links over to the algo column and then having the markers link to directly to the github source. The only problem i can think of here is that eg. "Item-Based Collaborative Filtering" has docs for both MapReduce and Spark, and in the future this may occur more often- though with the legacy status of MapReduce and the recent abstraction of CLI drivers, probably won't.
It would be nice to have another dimension to work in.
Thanks again,
Andy
> Date: Fri, 24 Oct 2014 10:54:09 +0200
> Subject: Re: Mahout 1.0 features (revisited)
> From: frank@frankscholten.nl
> To: dev@mahout.apache.org
>
> Great!
>
> On Fri, Oct 24, 2014 at 8:01 AM, Ted Dunning <te...@gmail.com> wrote:
>
> > On Thu, Oct 23, 2014 at 3:57 PM, Andrew Palumbo <ap...@outlook.com>
> > wrote:
> >
> > > Or I can just commit as is and people can have at the organization.
> > >
> >
> >
> > Sounds good to me!
> >
RE: Mahout 1.0 features (revisited)
Posted by Andrew Palumbo <ap...@outlook.com>.
Thank you guys for the feedback! I'll just commit the reorganized page in place of the current "List of Algorithms" page as is and the experts on each section can have at it / reorganize as they see fit.
I know that this page is not a big deal but my goal is to clarify- not confuse- so I really appreciate the eyes!
Another concern is that the page did not reflect the Hierarchical design of the new code.. ie. the Engine-bindings inheriting from and optimizing math-scala, which I think is a huge Mahout strong point. Maybe that is for another page...
Thanks for the comments, Thejas...
> > > I guess another thing that could be added would be (needs
> > > development) so that developer can dig in and start working.this is a good idea but might clutter the page, Instead maybe we can make it known that anything (non- MR) without a marker "needs development". Then maybe an explanation/link from the "How to Contribute page" regarding this fact.
> > > I think this is a very clear, and another thing we can add is maybe
> > > a link to the source code of the algorithm or an example/tutorial.I like this idea very much- maybe moving the doc links over to the algo column and then having the markers link to directly to the github source. The only problem i can think of here is that eg. "Item-Based Collaborative Filtering" has docs for both MapReduce and Spark, and in the future this may occur more often- though with the legacy status of MapReduce and the recent abstraction of CLI drivers, probably won't.
It would be nice to have another dimension to work in.
Thanks again,
Andy
> Date: Fri, 24 Oct 2014 10:54:09 +0200
> Subject: Re: Mahout 1.0 features (revisited)
> From: frank@frankscholten.nl
> To: dev@mahout.apache.org
>
> Great!
>
> On Fri, Oct 24, 2014 at 8:01 AM, Ted Dunning <te...@gmail.com> wrote:
>
> > On Thu, Oct 23, 2014 at 3:57 PM, Andrew Palumbo <ap...@outlook.com>
> > wrote:
> >
> > > Or I can just commit as is and people can have at the organization.
> > >
> >
> >
> > Sounds good to me!
> >
Re: Mahout 1.0 features (revisited)
Posted by Frank Scholten <fr...@frankscholten.nl>.
Great!
On Fri, Oct 24, 2014 at 8:01 AM, Ted Dunning <te...@gmail.com> wrote:
> On Thu, Oct 23, 2014 at 3:57 PM, Andrew Palumbo <ap...@outlook.com>
> wrote:
>
> > Or I can just commit as is and people can have at the organization.
> >
>
>
> Sounds good to me!
>
Re: Mahout 1.0 features (revisited)
Posted by Ted Dunning <te...@gmail.com>.
On Thu, Oct 23, 2014 at 3:57 PM, Andrew Palumbo <ap...@outlook.com> wrote:
> Or I can just commit as is and people can have at the organization.
>
Sounds good to me!
Re: Mahout 1.0 features (revisited)
Posted by thejas prasad <th...@gmail.com>.
Hey Andy,
I think this would be great! Especially for a user/ new dev this would be
awesome. I guess another thing that could be added would be (needs
development) so that developer can dig in and start working.
I think this is a very clear, and another thing we can add is maybe a link
to the source code of the algorithm or an example/tutorial.
Thanks,
Thejas
On Thu, Oct 23, 2014 at 3:57 PM, Andrew Palumbo <ap...@outlook.com> wrote:
> I made a list of Mahout 1.0 features by engine a while back which I’ve
> just reorganized (and added links to). I’m not sure whether It’s more
> confusing this way or the original way:
>
> Original:
>
> https://andrewpalumbo.github.io/algorithms_by_engine
>
> One concern is that it doesn’t emphasize the new work done in math-scala
> enough- given that there’s only one line of the table devoted to it. For
> this reason, I’ve reorganized it a bit.
>
> Reorganized:
>
> https://andrewpalumbo.github.io/features_by_engine
>
> But this becomes confusing- eg. having 2 entries for SSVD- one in “Mahout
> Math-Scala Core and Scala DSL” and one in “Collaborative Filtering with CLI
> Drivers”. For this reason I’ve tried to break it down into
> core-algorithms (math-scala, dsl, etc), and and end2end implementations of
> those algorithms (and others) with a CLI. Here though too, I feel that
> this is somewhat confusing: eg. does it seem that there is no SSVD
> algorithm implemented for Spark or H2O?
>
> I’d like to commit this in place of the “List of algorithms page”:
>
> http://mahout.apache.org/users/basics/algorithms.html
>
> This new table is inclusive of all that is on the list for 0.9 (with the
> exception of FPG).
>
> Briefly going over yesterday’s “Mahout vs Spark” thread:
> http://mail-archives.us.apache.org/mod_mbox/mahout-user/201410.mbox/browser,I’d
> like to put something out there to give users, developers and contributors
> a quick overview of what Mahout 1.0 is/has/does/extends/keeps/can be used
> for. However I don’t want it to cause any more confusion.
>
> There are probably some poorly described features, and maybe some
> incorrect links, so once I’ve committed it, you guys may want to have at it.
>
> Or I can just commit as is and people can have at the organization.
>
> I’m in favor of the revised/reorganized page. I’d appreciate feedback
> from any and all- specifically on whether the page is confusing and if so
> how it might be better organized.
>
> Thanks,
>
> Andy
>
>
Re: Mahout 1.0 features (revisited)
Posted by Mahesh Balija <ba...@gmail.com>.
Hi Andy,
That's really useful information for Mahout Dev/Users both.
Best,
Mahesh.B.
On Thu, Oct 23, 2014 at 11:57 PM, Andrew Palumbo <ap...@outlook.com> wrote:
> I made a list of Mahout 1.0 features by engine a while back which I’ve
> just reorganized (and added links to). I’m not sure whether It’s more
> confusing this way or the original way:
>
> Original:
>
> https://andrewpalumbo.github.io/algorithms_by_engine
>
> One concern is that it doesn’t emphasize the new work done in math-scala
> enough- given that there’s only one line of the table devoted to it. For
> this reason, I’ve reorganized it a bit.
>
> Reorganized:
>
> https://andrewpalumbo.github.io/features_by_engine
>
> But this becomes confusing- eg. having 2 entries for SSVD- one in “Mahout
> Math-Scala Core and Scala DSL” and one in “Collaborative Filtering with CLI
> Drivers”. For this reason I’ve tried to break it down into
> core-algorithms (math-scala, dsl, etc), and and end2end implementations of
> those algorithms (and others) with a CLI. Here though too, I feel that
> this is somewhat confusing: eg. does it seem that there is no SSVD
> algorithm implemented for Spark or H2O?
>
> I’d like to commit this in place of the “List of algorithms page”:
>
> http://mahout.apache.org/users/basics/algorithms.html
>
> This new table is inclusive of all that is on the list for 0.9 (with the
> exception of FPG).
>
> Briefly going over yesterday’s “Mahout vs Spark” thread:
> http://mail-archives.us.apache.org/mod_mbox/mahout-user/201410.mbox/browser,I’d
> like to put something out there to give users, developers and contributors
> a quick overview of what Mahout 1.0 is/has/does/extends/keeps/can be used
> for. However I don’t want it to cause any more confusion.
>
> There are probably some poorly described features, and maybe some
> incorrect links, so once I’ve committed it, you guys may want to have at it.
>
> Or I can just commit as is and people can have at the organization.
>
> I’m in favor of the revised/reorganized page. I’d appreciate feedback
> from any and all- specifically on whether the page is confusing and if so
> how it might be better organized.
>
> Thanks,
>
> Andy
>
>