and that only changes the fit by 0.01 between 1st degree and 2nd degree.
I want to use a weight vector (because this is the only other parameter I
can think of changing), but I can't figure out what to put in for the
vector. Help?
Olivia.
On Thu, 24 Oct 2002, Tao Li wrote:
> what neter books says is math. to pick the parameter for loess() is the
> same as picking parameter for any other funtions. you need to check the
> help(loess). the default seems to be one iteration. i have to see that
> most of the time people do not change the loess parameter. the default is
> usually good enough, which comes from complicated formula. and what do u
> mean by 'good?' if u know what is good in your opinion why do you need
> math formula!
>
> Tao Li
> --------------
> www.people.fas.harvard.edu/~li7
>
> On Thu, 24 Oct 2002, Olivia Lau wrote:
>
> > Neter p. 136 says that you should
> >
> > 1) use weights
> > 2) base the weights for the second iteration on the residuals for ths
> > first fitting.
> >
> > BUT, when I do summary(loess.object) it shows that the loess function only
> > does one iteration. Also, what should I use for my initial weight vector?
> > And my second weight vector?
> >
> > Thanks,
> >
> > Olivia
> >
> > _______________________________________________
> > gov1760-l mailing list
> > gov1760-l(a)fas.harvard.edu
> > http://www.fas.harvard.edu/mailman/listinfo/gov1760-l
> >
>
go to that directroy; start netscape; go to the class web page; go to the data
section; right click on the house data; say "save to file" and then netscape
will place the file in the current directory. You can then type "unzip
house.zip" to unzip the file.
You could also do all this from Windows and then just copy the files over.
Let us know if this doesn't get the job done (note that the above are mostly
correct but probably screw up some of the details).
Dave
aycock(a)fas.harvard.edu writes:
> This may sound like a very stupid question since i should have downloaded the
> data already and i should have already gotten a lot of work done on the problem
> set. Anyway, i have no idea how to download the data into my directory so i
> can't get started. Please help.
>
>
--
David Kane
Lecturer In Government
617-563-0122
dkane(a)latte.harvard.edu
In other words, you should type C-x C-c when you want to quit xemacs.
Matthew Cox via RT writes:
>
> I finally figured out what the users are doing that is leaving behind
> zombie xemacs processes.
>
> Instead of C-x,C-c to exit: they are using the X widgets and choosing
> destroy. Which destroys the window resource but leaves xemacs running
> without an xresource to draw to.
>
> Can you please advise the students the rpoper way to exit. They can even
> use the File menu. Just not the widgets in the window frame.
>
> Matt
>
They also note that:
> I was just helping one of the students in the lab. I think what they're
> doing wrong (or at least what this user was doing) was starting a shell
> inside xemacs and then running R from that shell.
> The result for the user I helped was that the help was being opened in
> Netscape. Which isn't bad I guess, but inside of another buffer is better.
Let me know if there are questions. To run R within xemacs, you type M-x R.
Dave
--
David Kane
Lecturer In Government
617-563-0122
dkane(a)latte.harvard.edu
can find nothing better. In retrospect, my key error was not assigning Moore
and McCabe (or an intro textbook of your choice) the first week of class. I
think that if you read chapter 3 of Moore and McCabe and then re-read QR 33, it
will make a lot more sense.
However, all of this weeks class with be R tricks and QR 33. So, please come
with many questions. I think that people did not ask enough questions about QR
33 last week . . . (do I get to give feedback too? --- ;-)
Thanks again for the feedback.
Dave
--
David Kane
Lecturer In Government
617-563-0122
dkane(a)latte.harvard.edu
and that only changes the fit by 0.01 between 1st degree and 2nd degree.
I want to use a weight vector (because this is the only other parameter I
can think of changing), but I can't figure out what to put in for the
vector. Help?
Olivia.
On Thu, 24 Oct 2002, Tao Li wrote:
> what neter books says is math. to pick the parameter for loess() is the
> same as picking parameter for any other funtions. you need to check the
> help(loess). the default seems to be one iteration. i have to see that
> most of the time people do not change the loess parameter. the default is
> usually good enough, which comes from complicated formula. and what do u
> mean by 'good?' if u know what is good in your opinion why do you need
> math formula!
>
> Tao Li
> --------------
> www.people.fas.harvard.edu/~li7
>
> On Thu, 24 Oct 2002, Olivia Lau wrote:
>
> > Neter p. 136 says that you should
> >
> > 1) use weights
> > 2) base the weights for the second iteration on the residuals for ths
> > first fitting.
> >
> > BUT, when I do summary(loess.object) it shows that the loess function only
> > does one iteration. Also, what should I use for my initial weight vector?
> > And my second weight vector?
> >
> > Thanks,
> >
> > Olivia
> >
> > _______________________________________________
> > gov1760-l mailing list
> > gov1760-l(a)fas.harvard.edu
> > http://www.fas.harvard.edu/mailman/listinfo/gov1760-l
> >
>
go to that directroy; start netscape; go to the class web page; go to the data
section; right click on the house data; say "save to file" and then netscape
will place the file in the current directory. You can then type "unzip
house.zip" to unzip the file.
You could also do all this from Windows and then just copy the files over.
Let us know if this doesn't get the job done (note that the above are mostly
correct but probably screw up some of the details).
Dave
aycock(a)fas.harvard.edu writes:
> This may sound like a very stupid question since i should have downloaded the
> data already and i should have already gotten a lot of work done on the problem
> set. Anyway, i have no idea how to download the data into my directory so i
> can't get started. Please help.
>
>
--
David Kane
Lecturer In Government
617-563-0122
dkane(a)latte.harvard.edu
In other words, you should type C-x C-c when you want to quit xemacs.
Matthew Cox via RT writes:
>
> I finally figured out what the users are doing that is leaving behind
> zombie xemacs processes.
>
> Instead of C-x,C-c to exit: they are using the X widgets and choosing
> destroy. Which destroys the window resource but leaves xemacs running
> without an xresource to draw to.
>
> Can you please advise the students the rpoper way to exit. They can even
> use the File menu. Just not the widgets in the window frame.
>
> Matt
>
They also note that:
> I was just helping one of the students in the lab. I think what they're
> doing wrong (or at least what this user was doing) was starting a shell
> inside xemacs and then running R from that shell.
> The result for the user I helped was that the help was being opened in
> Netscape. Which isn't bad I guess, but inside of another buffer is better.
Let me know if there are questions. To run R within xemacs, you type M-x R.
Dave
--
David Kane
Lecturer In Government
617-563-0122
dkane(a)latte.harvard.edu
can find nothing better. In retrospect, my key error was not assigning Moore
and McCabe (or an intro textbook of your choice) the first week of class. I
think that if you read chapter 3 of Moore and McCabe and then re-read QR 33, it
will make a lot more sense.
However, all of this weeks class with be R tricks and QR 33. So, please come
with many questions. I think that people did not ask enough questions about QR
33 last week . . . (do I get to give feedback too? --- ;-)
Thanks again for the feedback.
Dave
--
David Kane
Lecturer In Government
617-563-0122
dkane(a)latte.harvard.edu
and that only changes the fit by 0.01 between 1st degree and 2nd degree.
I want to use a weight vector (because this is the only other parameter I
can think of changing), but I can't figure out what to put in for the
vector. Help?
Olivia.
On Thu, 24 Oct 2002, Tao Li wrote:
> what neter books says is math. to pick the parameter for loess() is the
> same as picking parameter for any other funtions. you need to check the
> help(loess). the default seems to be one iteration. i have to see that
> most of the time people do not change the loess parameter. the default is
> usually good enough, which comes from complicated formula. and what do u
> mean by 'good?' if u know what is good in your opinion why do you need
> math formula!
>
> Tao Li
> --------------
> www.people.fas.harvard.edu/~li7
>
> On Thu, 24 Oct 2002, Olivia Lau wrote:
>
> > Neter p. 136 says that you should
> >
> > 1) use weights
> > 2) base the weights for the second iteration on the residuals for ths
> > first fitting.
> >
> > BUT, when I do summary(loess.object) it shows that the loess function only
> > does one iteration. Also, what should I use for my initial weight vector?
> > And my second weight vector?
> >
> > Thanks,
> >
> > Olivia
> >
> > _______________________________________________
> > gov1760-l mailing list
> > gov1760-l(a)fas.harvard.edu
> > http://www.fas.harvard.edu/mailman/listinfo/gov1760-l
> >
>
go to that directroy; start netscape; go to the class web page; go to the data
section; right click on the house data; say "save to file" and then netscape
will place the file in the current directory. You can then type "unzip
house.zip" to unzip the file.
You could also do all this from Windows and then just copy the files over.
Let us know if this doesn't get the job done (note that the above are mostly
correct but probably screw up some of the details).
Dave
aycock(a)fas.harvard.edu writes:
> This may sound like a very stupid question since i should have downloaded the
> data already and i should have already gotten a lot of work done on the problem
> set. Anyway, i have no idea how to download the data into my directory so i
> can't get started. Please help.
>
>
--
David Kane
Lecturer In Government
617-563-0122
dkane(a)latte.harvard.edu