Eric Cheney wrote:
On Sunday, 18 April 2004 at 0:52:19 +0200, Peter Dalgaard wrote:
Eric Cheney [EMAIL PROTECTED] writes:
Could someone tell me how to access the log likelihood
of a poisson model? I've done the following
BEGIN R STUFF
freq.mod - glm(formula = nfix ~ gls.gls + pol.gls
Dear James Wettenhall,
Your question - why do i need nonlinear regression for that model when it
is linear after taking logs - is not a dumb question: rather it is a
rational one. Actually C.E.S Production Function [ Y = GAMA *
((DELTA*K^(-BETA)) + ((1-DELTA)*L^(-BETA)))^(-PHI/BETA) ] is my
Dear Sundar Dorai-Raj,
Thank you very much for mentioning to exponentiate ALPHA.
However, so far i understand that the parameters in the non-linear equation
Y = ALPHA * (L^(BETA1)) * (K^(BETA2))
and the coefficients of log(L) and log(K) of the following equation (after
linearizing)
log(Y) =
It appears that the #If NeedFunctionPrototypes compiler directive has been
removed from Xlib.h and Xutil.h in Xfree86 4.4. All the prototypes
containing the offending _Xconst are now being processed. R 1.8.1, which
built successfully under XFree86 4.3, fails under XFree86 4.4 with the same
error
On Sun, 18 Apr 2004, Timothy Tatar wrote:
It appears that the #If NeedFunctionPrototypes compiler directive has been
removed from Xlib.h and Xutil.h in Xfree86 4.4. All the prototypes
containing the offending _Xconst are now being processed. R 1.8.1, which
built successfully under XFree86
Renaud Lancelot [EMAIL PROTECTED] writes:
And that's great; but I need the log likelihood.
Anyone know?
The deviance will not suffice? sum(dpois(nfix, fitted(freq.mod),
log.p=T))
should do the trick otherwise.
Thank you, that did the trick. I should note that the method
The short answer is no, as there is no way to recover the fitted values
and residuals so you can't get a proper fit object of class lm (and
hence get `summaries and all').
Your pseudo-data method needs to fix the u_i to be mean zero, variance
one in the sample. That is probably the quickest
Prof Brian Ripley [EMAIL PROTECTED] writes:
The short answer is no, as there is no way to recover the fitted values
and residuals so you can't get a proper fit object of class lm (and
hence get `summaries and all').
Your pseudo-data method needs to fix the u_i to be mean zero, variance
Thanks, Brian!
On 18-Apr-04 Prof Brian Ripley wrote:
The short answer is no, as there is no way to recover the fitted values
Well, the fitted values (a + b*x_i) would be available, as would be
the estimates and SEs of coefficients, sums of squares, and relevant
F ratios and P values.
and
Hello,
I routinely use aov and and the Error term to perform analyses of
variance of experiments with 'within-subject' factors. I wonder whether
a notion like 'multistratum models' exists for glm models when
performing a logit analysis (without being 100% sure whether this would
make sense).
Mohammad Ehsanul Karim wrote:
Dear Sundar Dorai-Raj,
Thank you very much for mentioning to exponentiate ALPHA.
However, so far i understand that the parameters in the non-linear equation
Y = ALPHA * (L^(BETA1)) * (K^(BETA2))
and the coefficients of log(L) and log(K) of the following equation
On Sun, 18 Apr 2004 [EMAIL PROTECTED] wrote:
Thanks, Brian!
On 18-Apr-04 Prof Brian Ripley wrote:
The short answer is no, as there is no way to recover the fitted values
Well, the fitted values (a + b*x_i) would be available, as would be
the estimates and SEs of coefficients, sums of
(Ted Harding) [EMAIL PROTECTED] writes:
Thanks, Brian!
On 18-Apr-04 Prof Brian Ripley wrote:
The short answer is no, as there is no way to recover the fitted values
Well, the fitted values (a + b*x_i) would be available, as would be
the estimates and SEs of coefficients, sums of
On 18-Apr-04 Peter Dalgaard wrote:
However, it begs the question whether it wouldn't have been better
to design the RSS into the lm class rather than computing it from
residuals in summary.lm and anova.lm and predict.lm and...
Well, something like this though lay under my original query.
The
On 18 Apr 2004 at 2:27, Ted Harding wrote:
Hi Folks,
I am dealing with data which have been presented as
at each x_i, mean m_i of the y-values at x_i,
sd s_i of the y-values at x_i
number n_i of the y-values at x_i
and I want to linearly regress y on x.
On 18-Apr-04 Peter Dalgaard wrote:
(Ted Harding) [EMAIL PROTECTED] writes:
On 18-Apr-04 Prof Brian Ripley wrote:
The short answer is no, as there is no way to recover the fitted
values
Well, the fitted values (a + b*x_i) would be available, as would be
the estimates and SEs of
As Doug had reported, Debian packages for R 1.9.0 were uploaded last Monday.
These are currently part of 'unstable', but can already be used on 'testing'
into which they should migrate in a few days.
Debian 'stable' is another matter. Neither Doug nor I has a stable system
left that would be
Hello Mateusz,
The 'hist' function works on the raw data.
In your data set example, you have already computed the number of data
points in each bin.
What you really want is probably a barplot of N
You could display your data:
plot(Class,N,'h')
Or
names(N)-Class
barplot(N)
Christophe Pallier
Hello,
Does anybody know if the outscale option of randomForest yields the
standarized version of the outlier measure for each case? or the results
are only the raw values. Also I have notice that this measure presents
very high variability. I mean if I repeat the experiment I am getting very
Hi Christophe,
On 4/18/2004 3:17 PM, Christophe Pallier wrote:
The 'hist' function works on the raw data.
In your data set example, you have already computed the number of data
points in each bin.
Yes, you are right. I evidently misunderstood the hist function
usage described in manuals.
What
The thing to do is probably:
1. Use fairly large number of trees (e.g., 1000).
2. Run a few times and average the results.
The reason for the instability is sort of two fold:
1. The random forest algorithm itself is based on randomization. That's why
it's probably a good idea to have 500-1000
I'm having some problems using variable names containing spaces (using
backticks) with gam (mgcv 0.9-6, R 1.8.1). Some toy code to reproduce
my problem is below. Am I doing something wrong, or should I pass this
bug on to Simon Wood? (Or do I need to rename my variables to get rid of
the
Hadley Wickham [EMAIL PROTECTED] writes:
I'm having some problems using variable names containing spaces (using
backticks) with gam (mgcv 0.9-6, R 1.8.1). Some toy code to reproduce
my problem is below. Am I doing something wrong, or should I pass
this bug on to Simon Wood? (Or do I need to
On Mon, 19 Apr 2004, Hadley Wickham wrote:
I'm having some problems using variable names containing spaces (using
backticks) with gam (mgcv 0.9-6, R 1.8.1). Some toy code to reproduce
my problem is below. Am I doing something wrong, or should I pass this
bug on to Simon Wood? (Or do I
Hi
jzhang10 wrote:
Hi,
I want to draw a level plot. The levels are not evenly spaced, so I did
something like: levels=c(0,2,5,10,30,60). I still want the color bar (key) on
the right side to be evenly spaced so that the small numbers (0,2,5) are not
squeezed together.
Does anyone know how to
Hi
I've a problem:
I wont to install R on a virtual server without root privileges.
I've tried to install in /usr/local/
but the fortran is not present.
I've tried to install libf2c and in the ./configure now is ok, but the
make crash.
I don't know if my libf2c is right...
Have a good link for
Given enough data, the choice between the two models can be made
in part by plotting the residuals vs. the predicted: or vs.
log(predicted): Suppose the true model was
log(Y) = log(ALPHA) +(BETA1)*log(L) + (BETA2)*log(K) + err,
where err is independent, normal with constant
Hello R users,
I am having difficulting getting multcomp to run.
I have a dataframe attached with a numeric variable q12a and a numeric variable quota
(which is really a classification variable).
quota has 10 levels and unequal sample sizes.
a12a has some missing data.
I am interested in
On Sun, 18 Apr 2004, Mohammad Ehsanul Karim wrote:
concern (In this case there is no way to linearize it), the Cobb-Douglas
being just a 'Toy problem' to see how non-linear process works. And i'm
sorry that i cannot guess some approximate parameter values for that CES
using some typical
On Sun, Apr 18, 2004 at 11:28:09PM -0400, Hector L. Ayala-del-Rio wrote:
Dear R-helpers,
I will like to know if there is a way to generate a stacked column
graph using both patterns and colors to fill the bars. I have many
categories for the number of color available in R, so I will
30 matches
Mail list logo
