You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@mahout.apache.org by Yoonmin Nam <ro...@dgist.ac.kr> on 2014/01/20 03:42:07 UTC
About Parallel Frequent Growth algorithm
Hello, everyone!
Is there anyone know about the reason why PFG is deprecated in Mahout?
I knew that new algorithm (BIGFIM) will be implemented as a substitution of
old PFG algorithm for parallel frequent pattern mining.
Please let me know if you knew the reason.
Thanks!
Re: About Parallel Frequent Growth algorithm
Posted by Dmitriy Lyubimov <dl...@gmail.com>.
it seems more like it is not supported.
I'd port it into spark counterpart and make sure there's a support (i.e. a
person to go after when it breaks :)
On Mon, Jan 20, 2014 at 3:44 PM, Suneel Marthi <su...@yahoo.com>wrote:
> I was asked this question too and I had no clear answer. May be it wasn't
> right to remove FP from the codebase.
> Not having this may well be one another reason for users to look at
> options other than Mahout.
>
> Given the issues that Frank's reported with Streaming KMeans (and I am
> seeing them too) I was gonna rollback the Release presently in staging
> anyways.
>
> Do we take a step back and restore FP for 0.9?
>
>
>
>
>
>
> On Monday, January 20, 2014 6:31 PM, Dmitriy Lyubimov <dl...@gmail.com>
> wrote:
>
> that's a bit weird though. Association mining is still a pretty popular
> technique. (our scientists use it, albeit not in exact FPGrowth form)
>
>
>
> On Mon, Jan 20, 2014 at 3:15 PM, Sebastian Schelter <ss...@apache.org>
> wrote:
>
> > Hi Yoonmin,
> >
> > we removed a bunch of algorithms either because they were rarely used or
> > not actively maintained anymore. IIRC the first thing was true for PFG.
> >
> > --sebastian
> >
> >
> > On 01/20/2014 03:42 AM, Yoonmin Nam wrote:
> >
> >> Hello, everyone!
> >>
> >>
> >>
> >> Is there anyone know about the reason why PFG is deprecated in Mahout?
> >>
> >>
> >>
> >> I knew that new algorithm (BIGFIM) will be implemented as a substitution
> >> of
> >> old PFG algorithm for parallel frequent pattern mining.
> >>
> >>
> >>
> >> Please let me know if you knew the reason.
> >>
> >>
> >>
> >> Thanks!
> >>
> >>
> >>
> >
>
Re: About Parallel Frequent Growth algorithm
Posted by Dmitriy Lyubimov <dl...@gmail.com>.
(1) diligently answer inqueries about method use and theoretical foundation
on the user list
(2) fix arising issues related to that stuff, diligently as well.
when (1) and (2) stops happening, the method gets axed in a year or couple
releases (which what pretty much happend this time i think).
You think you'd be able to subscribe for that for the next few years?
On Mon, Jan 20, 2014 at 4:46 PM, Qinghao Dai <ro...@gmail.com> wrote:
> May I ask what is the qualification to be the maintainer?
> I have read this part of code, and would like to have a try.
>
> Best Regards,
> Qinghao
>
>
> 2014/1/20 Ted Dunning <te...@gmail.com>
>
> > On Mon, Jan 20, 2014 at 5:44 PM, Suneel Marthi <suneel_marthi@yahoo.com
> > >wrote:
> >
> > > I was asked this question too and I had no clear answer. May be it
> wasn't
> > > right to remove FP from the codebase.
> > >
> >
> > The major problem was that we had no maintainers for the code.
> >
>
Re: About Parallel Frequent Growth algorithm
Posted by Qinghao Dai <ro...@gmail.com>.
May I ask what is the qualification to be the maintainer?
I have read this part of code, and would like to have a try.
Best Regards,
Qinghao
2014/1/20 Ted Dunning <te...@gmail.com>
> On Mon, Jan 20, 2014 at 5:44 PM, Suneel Marthi <suneel_marthi@yahoo.com
> >wrote:
>
> > I was asked this question too and I had no clear answer. May be it wasn't
> > right to remove FP from the codebase.
> >
>
> The major problem was that we had no maintainers for the code.
>
Re: About Parallel Frequent Growth algorithm
Posted by Ted Dunning <te...@gmail.com>.
On Mon, Jan 20, 2014 at 5:44 PM, Suneel Marthi <su...@yahoo.com>wrote:
> I was asked this question too and I had no clear answer. May be it wasn't
> right to remove FP from the codebase.
>
The major problem was that we had no maintainers for the code.
Re: About Parallel Frequent Growth algorithm
Posted by Suneel Marthi <su...@yahoo.com>.
I was asked this question too and I had no clear answer. May be it wasn't right to remove FP from the codebase.
Not having this may well be one another reason for users to look at options other than Mahout.
Given the issues that Frank's reported with Streaming KMeans (and I am seeing them too) I was gonna rollback the Release presently in staging anyways.
Do we take a step back and restore FP for 0.9?
On Monday, January 20, 2014 6:31 PM, Dmitriy Lyubimov <dl...@gmail.com> wrote:
that's a bit weird though. Association mining is still a pretty popular
technique. (our scientists use it, albeit not in exact FPGrowth form)
On Mon, Jan 20, 2014 at 3:15 PM, Sebastian Schelter <ss...@apache.org> wrote:
> Hi Yoonmin,
>
> we removed a bunch of algorithms either because they were rarely used or
> not actively maintained anymore. IIRC the first thing was true for PFG.
>
> --sebastian
>
>
> On 01/20/2014 03:42 AM, Yoonmin Nam wrote:
>
>> Hello, everyone!
>>
>>
>>
>> Is there anyone know about the reason why PFG is deprecated in Mahout?
>>
>>
>>
>> I knew that new algorithm (BIGFIM) will be implemented as a substitution
>> of
>> old PFG algorithm for parallel frequent pattern mining.
>>
>>
>>
>> Please let me know if you knew the reason.
>>
>>
>>
>> Thanks!
>>
>>
>>
>
Re: About Parallel Frequent Growth algorithm
Posted by Dmitriy Lyubimov <dl...@gmail.com>.
that's a bit weird though. Association mining is still a pretty popular
technique. (our scientists use it, albeit not in exact FPGrowth form)
On Mon, Jan 20, 2014 at 3:15 PM, Sebastian Schelter <ss...@apache.org> wrote:
> Hi Yoonmin,
>
> we removed a bunch of algorithms either because they were rarely used or
> not actively maintained anymore. IIRC the first thing was true for PFG.
>
> --sebastian
>
>
> On 01/20/2014 03:42 AM, Yoonmin Nam wrote:
>
>> Hello, everyone!
>>
>>
>>
>> Is there anyone know about the reason why PFG is deprecated in Mahout?
>>
>>
>>
>> I knew that new algorithm (BIGFIM) will be implemented as a substitution
>> of
>> old PFG algorithm for parallel frequent pattern mining.
>>
>>
>>
>> Please let me know if you knew the reason.
>>
>>
>>
>> Thanks!
>>
>>
>>
>
Re: About Parallel Frequent Growth algorithm
Posted by Sebastian Schelter <ss...@apache.org>.
Hi Yoonmin,
we removed a bunch of algorithms either because they were rarely used or
not actively maintained anymore. IIRC the first thing was true for PFG.
--sebastian
On 01/20/2014 03:42 AM, Yoonmin Nam wrote:
> Hello, everyone!
>
>
>
> Is there anyone know about the reason why PFG is deprecated in Mahout?
>
>
>
> I knew that new algorithm (BIGFIM) will be implemented as a substitution of
> old PFG algorithm for parallel frequent pattern mining.
>
>
>
> Please let me know if you knew the reason.
>
>
>
> Thanks!
>
>