Thanks Andy.
I am reading the "locfit" document...
but not sure how to do the CV and bandwidth selection...
Here is a quote about the function "regband": it doesn't seem to be usable?
Basically I am looking for a "locfit" that comes with an automatic
bandwidth selection so that I am essentially parameter free for the
local-regression step...
-------------------------------------
regband Bandwidth selectors for local regression.
Description
Function to compute local regression bandwidths for local linear
regression, implemented as a front
end to locfit().
This function is included for comparative purposes only. Plug-in selectors
are based on flawed logic,
make unreasonable and restrictive assumptions and do not use the full power
of the estimates available
in Locfit. Any relation between the results produced by this function and
desirable estimates
are entirely coincidental.
Usage
regband(formula, what = c("CP", "GCV", "GKK", "RSW"), deg=1, ...)
2012/2/23 Liaw, Andy <andy_l...@merck.com>
> **
> If that's the kind of framework you'd like to work in, use locfit, which
> has the predict() method for evaluating new data. There are several
> different handwidth selectors in that package for your choosing.
>
> Kernel smoothers don't really fit the framework of "creating a model
> object, followed by predicting new data using that fitted model object"
> very well because of it's local nature. Think of k-nn classification,
> which has similar problem: The "model" needs to be computed for every data
> point you want to predict.
>
> Andy
>
> ------------------------------
> *From:* Michael [mailto:comtech....@gmail.com]
> *Sent:* Thursday, February 23, 2012 10:06 AM
>
> *To:* Liaw, Andy
> *Cc:* Bert Gunter; r-help
> *Subject:* Re: [R] Good and modern Kernel Regression package in R with
> auto-bandwidth?
>
> Thank you Andy!
>
> I went thru KernSmooth package but I don't see a way to use the fitted
> function to do the "predict" part...
>
>
> data=data.frame(z=z, x=x)
>
> datanew=data.frame(z=z, x=x)
>
> lmfit=lm(z
> ~x, data=data)
>
> lmforecast=predict(lmfit, newdata=datanew)
>
> Am I missing anything here?
>
> Thanks!
> 2012/2/23 Liaw, Andy <andy_l...@merck.com>
>
>> **
>> In short, pick your poison...
>>
>> Is there any particular reason why the tools that shipped with R itself
>> (e.g., kernSmooth) are inadequate for you?
>>
>> I like using the locfit package because it has many tools, including the
>> ones that the author didn't think were optimal. You may need the book to
>> get most mileage out of it though.
>>
>> Andy
>>
>> ------------------------------
>> *From:* Michael [mailto:comtech....@gmail.com]
>> *Sent:* Thursday, February 23, 2012 12:25 AM
>> *To:* Liaw, Andy
>> *Cc:* Bert Gunter; r-help
>>
>> *Subject:* Re: [R] Good and modern Kernel Regression package in R with
>> auto-bandwidth?
>>
>> Ｉmeant its very slow when I use "cv.aic"...
>>
>> On Wed, Feb 22, 2012 at 11:24 PM, Michael <comtech....@gmail.com> wrote:
>>
>>> Is "np" an okay package to use?
>>>
>>> I am worried about the "multi-start" thing... and also it's very slow...
>>>
>>>
>>> On Wed, Feb 22, 2012 at 8:35 PM, Liaw, Andy <andy_l...@merck.com> wrote:
>>>
>>>> Bert's question aside (I was going to ask about laundry, but that's
>>>> much harder than taxes...), my understanding of the situation is that
>>>> "optimal" is in the eye of the beholder. There were at least two schools
>>>> of thought on which is the better way of automatically selecting bandwidth,
>>>> using plug-in methods or CV-type. The last I check, the jury is still out.
>>>>
>>>> Andy
>>>>
>>>> > -----Original Message-----
>>>> > From: r-help-boun...@r-project.org
>>>> > [mailto:r-help-boun...@r-project.org] On Behalf Of Bert Gunter
>>>> > Sent: Wednesday, February 22, 2012 6:03 PM
>>>> > To: Michael
>>>> > Cc: r-help
>>>> > Subject: Re: [R] Good and modern Kernel Regression package in
>>>> > R with auto-bandwidth?
>>>> >
>>>> > Would you like it to do your your taxes for you too? :-)
>>>> >
>>>> > Bert
>>>> >
>>>> > Sent from my iPhone -- please excuse typos.
>>>> >
>>>> > On Feb 22, 2012, at 11:46 AM, Michael <comtech....@gmail.com> wrote:
>>>> >
>>>> > > Hi all,
>>>> > >
>>>> > > I am looking for a good and modern Kernel Regression
>>>> > package in R, which
>>>> > > has the following features:
>>>> > >
>>>> > > 1) It has cross-validation
>>>> > > 2) It can automatically choose the "optimal" bandwidth
>>>> > > 3) It doesn't have random effect - i.e. if I run the
>>>> > function at different
>>>> > > times on the same data-set, the results should be exactly
>>>> > the same... I am
>>>> > > trying "np", but I am seeing:
>>>> > >
>>>> > > Multistart 1 of 1 |
>>>> > > Multistart 1 of 1 |
>>>> > > ...
>>>> > >
>>>> > > It looks like in order to do the optimization, it's doing
>>>> > > multiple-random-start optimization... am I right?
>>>> > >
>>>> > >
>>>> > > Could you please give me some pointers?
>>>> > >
>>>> > > I did some google search but there are so many packages
>>>> > that do this... I
>>>> > > just wanted to find the best/modern one to use...
>>>> > >
>>>> > > Thank you!
>>>> > >
>>>> > > [[alternative HTML version deleted]]
>>>> > >
>>>> > > ______________________________________________
>>>> > > R-help@r-project.org mailing list
>>>> > > https://stat.ethz.ch/mailman/listinfo/r-help
>>>> > > PLEASE do read the posting guide
>>>> > http://www.R-project.org/posting-guide.html<http://www.r-project.org/posting-guide.html>
>>>> > > and provide commented, minimal, self-contained, reproducible code.
>>>> >
>>>> > ______________________________________________
>>>> > R-help@r-project.org mailing list
>>>> > https://stat.ethz.ch/mailman/listinfo/r-help
>>>> > PLEASE do read the posting guide
>>>> > http://www.R-project.org/posting-guide.html<http://www.r-project.org/posting-guide.html>
>>>> > and provide commented, minimal, self-contained, reproducible code.
>>>> >
>>>> Notice: This e-mail message, together with any attachments, contains
>>>> information of Merck & Co., Inc. (One Merck Drive, Whitehouse Station,
>>>> New Jersey, USA 08889), and/or its affiliates Direct contact information
>>>> for affiliates is available at
>>>> http://www.merck.com/contact/contacts.html) that may be confidential,
>>>> proprietary copyrighted and/or legally privileged. It is intended solely
>>>> for the use of the individual or entity named on this message. If you
>>>> are
>>>> not the intended recipient, and have received this message in error,
>>>> please notify us immediately by reply e-mail and then delete it from
>>>> your system.
>>>>
>>>>
>>>
>> Notice: This e-mail message, together with any attachments, contains
>> information of Merck & Co., Inc. (One Merck Drive, Whitehouse Station,
>> New Jersey, USA 08889), and/or its affiliates Direct contact information
>> for affiliates is available at
>> http://www.merck.com/contact/contacts.html) that may be confidential,
>> proprietary copyrighted and/or legally privileged. It is intended solely
>> for the use of the individual or entity named on this message. If you are
>> not the intended recipient, and have received this message in error,
>> please notify us immediately by reply e-mail and then delete it from
>> your system.
>>
>
> Notice: This e-mail message, together with any attachments, contains
> information of Merck & Co., Inc. (One Merck Drive, Whitehouse Station,
> New Jersey, USA 08889), and/or its affiliates Direct contact information
> for affiliates is available at
> http://www.merck.com/contact/contacts.html) that may be confidential,
> proprietary copyrighted and/or legally privileged. It is intended solely
> for the use of the individual or entity named on this message. If you are
> not the intended recipient, and have received this message in error,
> please notify us immediately by reply e-mail and then delete it from
> your system.
>
[[alternative HTML version deleted]]
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
