You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@maven.apache.org by Benson Margulies <bi...@gmail.com> on 2011/04/29 13:30:42 UTC
Logistic Regression
With some help from Ted (which I plan to turn into a checked-in tool
if he doesn't get there first), I'm running LR on my initial small
example.
I adapted Ted's rcv1 sample to digest a directory containing
subdirectories containing exemplars.
Ted's delightfully small program pushes all of the data into the model
'n' times (n is 10 in my current variation. It displays the best
learner's accuracy at each iteration.
The example is 1000 docs in 10 categories.
With 20k-features, I note that the accuracy scores get worse on each
iteration of pushing the data into the model.
After the first pass, the model hasn't trained yet. After the second,
accuracy is 95.6%, and then if drifts gracefully downward with each
additional iteration, landing at .83.
I'm puzzled; I'm accustomed to overfitting causing scores to inflate,
but this pattern is not intuitive to me.
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@maven.apache.org
For additional commands, e-mail: users-help@maven.apache.org
Re: Logistic Regression
Posted by Benson Margulies <bi...@gmail.com>.
This went to the wrong list. Apologies.
On Fri, Apr 29, 2011 at 7:30 AM, Benson Margulies <bi...@gmail.com> wrote:
> With some help from Ted (which I plan to turn into a checked-in tool
> if he doesn't get there first), I'm running LR on my initial small
> example.
>
> I adapted Ted's rcv1 sample to digest a directory containing
> subdirectories containing exemplars.
>
> Ted's delightfully small program pushes all of the data into the model
> 'n' times (n is 10 in my current variation. It displays the best
> learner's accuracy at each iteration.
>
> The example is 1000 docs in 10 categories.
>
> With 20k-features, I note that the accuracy scores get worse on each
> iteration of pushing the data into the model.
>
> After the first pass, the model hasn't trained yet. After the second,
> accuracy is 95.6%, and then if drifts gracefully downward with each
> additional iteration, landing at .83.
>
> I'm puzzled; I'm accustomed to overfitting causing scores to inflate,
> but this pattern is not intuitive to me.
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@maven.apache.org
For additional commands, e-mail: users-help@maven.apache.org
Re: Logistic Regression
Posted by Hervé BOUTEMY <he...@free.fr>.
just for the record:
Maven and Mahout are Apache projects
Mojo and M2E aren't Apache projects, but only Maven-related, each one with its
own governance
Regards,
Hervé
Le vendredi 29 avril 2011, Benson Margulies a écrit :
> Wendy got it in one.
>
> Mahout, Maven, Mojo, M2E. One slight slip of the fingers and the wrong
> one ends up in the gmail to box.
>
> On Fri, Apr 29, 2011 at 5:06 PM, Wendy Smoak <ws...@gmail.com> wrote:
> > On Fri, Apr 29, 2011 at 4:44 PM, Wayne Fay <wa...@gmail.com> wrote:
> >>> I did send a message along afterwards apologizing. Did that not get
> >>> through?
> >>
> >> I got it and assume everyone else did too.
> >>
> >>> Too many apache projects start with 'm'.
> >>
> >> Simply to satisfy my curiousity, which project that starts with "m"
> >> were you talking about in your original email? ;-)
> >
> > I'm going to guess... Mahout. :)
> >
> > --
> > Wendy
> >
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: users-unsubscribe@maven.apache.org
> > For additional commands, e-mail: users-help@maven.apache.org
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@maven.apache.org
> For additional commands, e-mail: users-help@maven.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@maven.apache.org
For additional commands, e-mail: users-help@maven.apache.org
Re: Logistic Regression
Posted by Benson Margulies <bi...@gmail.com>.
Wendy got it in one.
Mahout, Maven, Mojo, M2E. One slight slip of the fingers and the wrong
one ends up in the gmail to box.
On Fri, Apr 29, 2011 at 5:06 PM, Wendy Smoak <ws...@gmail.com> wrote:
> On Fri, Apr 29, 2011 at 4:44 PM, Wayne Fay <wa...@gmail.com> wrote:
>>> I did send a message along afterwards apologizing. Did that not get through?
>>
>> I got it and assume everyone else did too.
>>
>>> Too many apache projects start with 'm'.
>>
>> Simply to satisfy my curiousity, which project that starts with "m"
>> were you talking about in your original email? ;-)
>
> I'm going to guess... Mahout. :)
>
> --
> Wendy
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@maven.apache.org
> For additional commands, e-mail: users-help@maven.apache.org
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@maven.apache.org
For additional commands, e-mail: users-help@maven.apache.org
Re: Logistic Regression
Posted by Wendy Smoak <ws...@gmail.com>.
On Fri, Apr 29, 2011 at 4:44 PM, Wayne Fay <wa...@gmail.com> wrote:
>> I did send a message along afterwards apologizing. Did that not get through?
>
> I got it and assume everyone else did too.
>
>> Too many apache projects start with 'm'.
>
> Simply to satisfy my curiousity, which project that starts with "m"
> were you talking about in your original email? ;-)
I'm going to guess... Mahout. :)
--
Wendy
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@maven.apache.org
For additional commands, e-mail: users-help@maven.apache.org
Re: Logistic Regression
Posted by Wayne Fay <wa...@gmail.com>.
> I did send a message along afterwards apologizing. Did that not get through?
I got it and assume everyone else did too.
> Too many apache projects start with 'm'.
Simply to satisfy my curiousity, which project that starts with "m"
were you talking about in your original email? ;-)
Wayne
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@maven.apache.org
For additional commands, e-mail: users-help@maven.apache.org
Re: Logistic Regression
Posted by Benson Margulies <bi...@gmail.com>.
Anders,
I did send a message along afterwards apologizing. Did that not get through?
Too many apache projects start with 'm'.
--benson
On Fri, Apr 29, 2011 at 7:37 AM, Anders Hammar <an...@hammar.net> wrote:
> Re-reading your mail a few times I seriously hope that this was posted to
> the wrong list, as I don't understand anything...?
>
> /Anders
>
> On Fri, Apr 29, 2011 at 13:30, Benson Margulies <bi...@gmail.com>wrote:
>
>> With some help from Ted (which I plan to turn into a checked-in tool
>> if he doesn't get there first), I'm running LR on my initial small
>> example.
>>
>> I adapted Ted's rcv1 sample to digest a directory containing
>> subdirectories containing exemplars.
>>
>> Ted's delightfully small program pushes all of the data into the model
>> 'n' times (n is 10 in my current variation. It displays the best
>> learner's accuracy at each iteration.
>>
>> The example is 1000 docs in 10 categories.
>>
>> With 20k-features, I note that the accuracy scores get worse on each
>> iteration of pushing the data into the model.
>>
>> After the first pass, the model hasn't trained yet. After the second,
>> accuracy is 95.6%, and then if drifts gracefully downward with each
>> additional iteration, landing at .83.
>>
>> I'm puzzled; I'm accustomed to overfitting causing scores to inflate,
>> but this pattern is not intuitive to me.
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: users-unsubscribe@maven.apache.org
>> For additional commands, e-mail: users-help@maven.apache.org
>>
>>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@maven.apache.org
For additional commands, e-mail: users-help@maven.apache.org
Re: Logistic Regression
Posted by Anders Hammar <an...@hammar.net>.
Re-reading your mail a few times I seriously hope that this was posted to
the wrong list, as I don't understand anything...?
/Anders
On Fri, Apr 29, 2011 at 13:30, Benson Margulies <bi...@gmail.com>wrote:
> With some help from Ted (which I plan to turn into a checked-in tool
> if he doesn't get there first), I'm running LR on my initial small
> example.
>
> I adapted Ted's rcv1 sample to digest a directory containing
> subdirectories containing exemplars.
>
> Ted's delightfully small program pushes all of the data into the model
> 'n' times (n is 10 in my current variation. It displays the best
> learner's accuracy at each iteration.
>
> The example is 1000 docs in 10 categories.
>
> With 20k-features, I note that the accuracy scores get worse on each
> iteration of pushing the data into the model.
>
> After the first pass, the model hasn't trained yet. After the second,
> accuracy is 95.6%, and then if drifts gracefully downward with each
> additional iteration, landing at .83.
>
> I'm puzzled; I'm accustomed to overfitting causing scores to inflate,
> but this pattern is not intuitive to me.
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@maven.apache.org
> For additional commands, e-mail: users-help@maven.apache.org
>
>