You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@mahout.apache.org by Ted Dunning <te...@gmail.com> on 2013/07/01 10:22:22 UTC
Re: mahout logistic regression
Follow into the regression code itself and check the references.
On Fri, Jun 28, 2013 at 3:35 PM, qiaoresearcher <qi...@gmail.com>wrote:
> The logistic regression code is difficult to follow: the trainlogistic and
> runlogistic part
>
> how the likelihood is calculated, how the weights is updated, etc
>
> does anyone knows who write the mahout logistic regression code? what are
> the reference on logistic regression algorithm he was using to write the
> code?
>
> thanks,
>
Re: mahout logistic regression
Posted by qiaoresearcher <qi...@gmail.com>.
Thanks for the paper you sent. It helps.
On Mon, Jul 1, 2013 at 3:00 PM, Ted Dunning <te...@gmail.com> wrote:
> I believe that there is a link to this paper in the code. If not file a
> jira.
>
>
> http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.177.3514&rep=rep1&type=pdf
>
> Note also that stochastic gradient descent is a very common algorithm for
> large scale logistic regression. You can find the basics anywhere with a
> simple google search.
>
>
> Sent from my iPhone
>
>
> On Jul 1, 2013, at 11:59, qiaoresearcher <qi...@gmail.com> wrote:
>
> > Ted,
> >
> > Thanks, but I have looked into the code and found it not very clear to
> me.
> > For a given cost function, there are many ways to optimize the cost
> > function, eg, gradient descent, newton method. It would be much better if
> > the authors of Mahout code can simply put one line in some place to say:
> > the implementation is based on XXX paper or XXX book chapter or XXX
> > webpage. Otherwise it is not easy for the users to figure out what update
> > rule is used in the code...
> >
> > Regards,
> >
> > On Mon, Jul 1, 2013 at 3:22 AM, Ted Dunning <te...@gmail.com>
> wrote:
> >
> >> Follow into the regression code itself and check the references.
> >>
> >>
> >> On Fri, Jun 28, 2013 at 3:35 PM, qiaoresearcher <
> qiaoresearcher@gmail.com
> >>> wrote:
> >>
> >>> The logistic regression code is difficult to follow: the trainlogistic
> >> and
> >>> runlogistic part
> >>>
> >>> how the likelihood is calculated, how the weights is updated, etc
> >>>
> >>> does anyone knows who write the mahout logistic regression code? what
> are
> >>> the reference on logistic regression algorithm he was using to write
> the
> >>> code?
> >>>
> >>> thanks,
> >>
>
Re: mahout logistic regression
Posted by Ted Dunning <te...@gmail.com>.
I believe that there is a link to this paper in the code. If not file a jira.
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.177.3514&rep=rep1&type=pdf
Note also that stochastic gradient descent is a very common algorithm for large scale logistic regression. You can find the basics anywhere with a simple google search.
Sent from my iPhone
On Jul 1, 2013, at 11:59, qiaoresearcher <qi...@gmail.com> wrote:
> Ted,
>
> Thanks, but I have looked into the code and found it not very clear to me.
> For a given cost function, there are many ways to optimize the cost
> function, eg, gradient descent, newton method. It would be much better if
> the authors of Mahout code can simply put one line in some place to say:
> the implementation is based on XXX paper or XXX book chapter or XXX
> webpage. Otherwise it is not easy for the users to figure out what update
> rule is used in the code...
>
> Regards,
>
> On Mon, Jul 1, 2013 at 3:22 AM, Ted Dunning <te...@gmail.com> wrote:
>
>> Follow into the regression code itself and check the references.
>>
>>
>> On Fri, Jun 28, 2013 at 3:35 PM, qiaoresearcher <qiaoresearcher@gmail.com
>>> wrote:
>>
>>> The logistic regression code is difficult to follow: the trainlogistic
>> and
>>> runlogistic part
>>>
>>> how the likelihood is calculated, how the weights is updated, etc
>>>
>>> does anyone knows who write the mahout logistic regression code? what are
>>> the reference on logistic regression algorithm he was using to write the
>>> code?
>>>
>>> thanks,
>>
Re: mahout logistic regression
Posted by qiaoresearcher <qi...@gmail.com>.
Ted,
Thanks, but I have looked into the code and found it not very clear to me.
For a given cost function, there are many ways to optimize the cost
function, eg, gradient descent, newton method. It would be much better if
the authors of Mahout code can simply put one line in some place to say:
the implementation is based on XXX paper or XXX book chapter or XXX
webpage. Otherwise it is not easy for the users to figure out what update
rule is used in the code...
Regards,
On Mon, Jul 1, 2013 at 3:22 AM, Ted Dunning <te...@gmail.com> wrote:
> Follow into the regression code itself and check the references.
>
>
> On Fri, Jun 28, 2013 at 3:35 PM, qiaoresearcher <qiaoresearcher@gmail.com
> >wrote:
>
> > The logistic regression code is difficult to follow: the trainlogistic
> and
> > runlogistic part
> >
> > how the likelihood is calculated, how the weights is updated, etc
> >
> > does anyone knows who write the mahout logistic regression code? what are
> > the reference on logistic regression algorithm he was using to write the
> > code?
> >
> > thanks,
> >
>