gmane.comp.lang.r.general

glmnet object to pmml

yan — 2012-05-18T12:23:04

Dear R users

I generated a model using glmnet, I need to convert it to pmml, but R pmml
package doesn't support glmnet, has anyone come across similar problem? any
idea to solve it?

Many thanks

YAn

--
View this message in context: http://r.789695.n4.nabble.com/glmnet-object-to-pmml-tp4630493.html
Sent from the R help mailing list archive at Nabble.com.

Re: Complex sort problem

Petr Savicky — 2012-05-18T11:48:58

The suggestion

  set.seed(12345)
  x <- sample(0:100, 10)
  x.order <- order(x)
  x.sorted <- x[x.order]

  sample.ind <- sample(1:length(x), 5, replace = TRUE)  #sample 1/2 size with replacement

  x.sample <- x.sorted[sample.ind]
  freq <- tabulate(sample.ind, nbins=length(x))
  x.sample.sorted <- rep(x.sorted, times=freq)

uses the fact that rep(x.sorted, times=freq) keeps the order
in x.sorted. This x.sorted can be a data frame, in which
case we should use 

  sample.ind <- sample(1:nrow(x), 5, replace = TRUE)
  x.sample <- x.sorted[sample.ind, ]
  freq <- tabulate(sample.ind, nbins=nrow(x))
  x.sample.sorted <- x.sorted[rep(1:nrow(x.sorted), times=freq), ]

It is possible to have several x.sorted data frames sorted according
to different variables. In this case, we generate pairs x.sample and
x.sample.sorted which are the same sample once unsorted and once sorted.
However, we get different samples for each sorting variable.

In order to save CPU time, if the same sample should be sortable
by different

Re: MANOVA with random factor

Helios de Rosario — 2012-05-18T10:53:59

I suppose you mean this:
http://www.r-project.org/doc/Rnews/Rnews_2007-2.pdf

Another approach, less general but in my opinion easier to use (and
equivalent in many situations), is provided by Anova() in package car.
See:
http://socserv.mcmaster.ca/jfox/Books/Companion/appendix/Appendix-Multivariate-Linear-Models.pdf

The package ez also has ezANOVA(), a wrapper to Anova() that may
simplify the task for full factorial designs.

Helios De Rosario



<David.Costantini< at >glasgow.ac.uk> escribió:
group*sex) and
does 

INSTITUTO DE BIOMECÁNICA DE VALENCIA
Universidad Politécnica de Valencia • Edificio 9C
Camino de Vera s/n • 46022 VALENCIA (ESPAÑA)
Tel. +34 96 387 91 60 • Fax +34 96 387 91 69
www.ibv.org

  Antes de imprimir este e-mail piense bien si es necesario hacerlo.
En cumplimiento de la Ley Orgánica 15/1999 reguladora de la Protección
de Datos de Carácter Personal, le informamos de que el presente mensaje
contiene información confidencial, siendo para uso exclusivo del
destinatario arriba indi

Re: Complex sort problem

Axel Urbiz — 2012-05-18T10:37:03

Would I be able to accomplish the same if x.sample was created from x
instead of x.sorted. The problem is that in my real problem, I have to sort
with respect to many variables and thus keep the sample indexes consistent
across variables. So I need to first take the sample and then sort it
with respect to potentially any variable.

Thanks again,
Axel.

On Thu, May 17, 2012 at 1:43 PM, Petr Savicky <savicky< at >cs.cas.cz> wrote:


[[alternative HTML version deleted]]

Re: Financial Statements Date Subsetting

R. Michael Weylandt — 2012-05-18T10:26:03

You're leaving out a step -- take a look at the examples for viewFin
and note that you have to download the financial data first with
getFin()

Michael

On Thu, May 17, 2012 at 9:53 PM, Brian Edmundson
<brianedmundson1< at >gmail.com> wrote:

Re: Proc AutoReg (SAS like Output in R)

R. Michael Weylandt — 2012-05-18T10:23:54

Well, the easiest way would be to use SAS ;-)

In R take a look at the ar() function. You can view documentation by
typing ?ar at the console and see some worked examples by running
example(ar)

Hope this helps,
Michael

On Fri, May 18, 2012 at 12:11 AM, anil <anilshivkumar< at >gmail.com> wrote:

Re: Optimization inconsistencies

Petr Savicky — 2012-05-18T09:42:43

Hi.

As Peter Dalgaard suggested, lpSolve may be used. If "low" may contain
negative values, then it is important to note that lpSolve assumes
all variables nonnegative. So, a transformation is needed, for example
x = low + y and solve for y. The following is one approach.

  library(lpSolve)
 
  n <- 8
  w <- 1:n
  low <- rep(-1, times=n)
  high <- rep(1, times=n)
 
  crit <- w
  mat <- rbind(diag(n), 1)
  rhs <- c(high - low, 1 - sum(low))
  dir <- c(rep("<=", times=n), "==")
 
  out <- lp("max", objective.in=crit, const.mat=mat, const.dir=dir, const.rhs=rhs)
  x <- low + out$solution

  round(x, digits=15)

  [1] -1 -1 -1  0  1  1  1  1

Hope this helps.

Petr Savicky.

Natural Language Toolkit for R - is it still there?

Nick Riches — 2012-05-18T09:41:43

Hi

I used the package Natural Language Toolkit about a year ago to parse
strings into parts of speech. However, I cannot find it on CRAN.

Has it been withdrawn?

Thanks

Nick

Re: Optimization inconsistencies

Hans W Borchers — 2012-05-18T08:33:49


Same as before:
Would you *please* post data and code, as the posting guide requests!

The variable that is left out still has some (linear) inequalities to fullfil. 
I didn't understand how you worked that out with optim() as it only allows to
include box constraints.

Hans Werner


______________________________________________
R-help< at >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.

Re: Optimization inconsistencies

peter dalgaard — 2012-05-18T08:31:45

On May 18, 2012, at 09:10 , Hans W Borchers wrote:


I considered making a similar remark, then realized that lpSolve actually allows equality constraints, so why not just use the tool that is designed for the job?



However, with a linear objective function, the Hessian is 0 and the maximum is attained at a corner point, which is likely to confuse algorithms that expect a locally quadratic function. 


Yes, also from the development perspective. We need to see more of these hard examples.

High Frequency Data

rahul deora — 2012-05-18T07:52:59

Is there any way to analyse high frequency data in R like
cleaning,manipulation and volatility etc.I know there are packages like
RTAQ and Realized for analysing high frequency data but they are only valid
for NYSE stocks and have well defined data format.

Please help.

[[alternative HTML version deleted]]

Re: Optimization inconsistencies

Marc Girondot — 2012-05-18T07:38:09

Le 18/05/12 00:14, Nathan Stephens a écrit :
I had similar problem to solve (x were frequencies) and optimization 
stops before to reach the global maximum. As hwborchers< at >googlemail.com 
indicates, I fitted {x}-1 values because the last one is known by the 
equality constraint.
For the vector constraints, I used w <- -Inf when x goes out of the limits.

Finally I used Bayesian mcmc to get the convergence and it works much 
better.

I don't know why in this case the optim does not converge.

Hope it hepls,

Marc

Re: glm convergence warning

Prof Brian Ripley — 2012-05-18T07:36:06

Actually, no, it did not.  The actual message is

 > Warning: glm.fit: algorithm did not converge

Had you shown us that rather than misquote it, all would have been clearer.

warning "algorithm did not converge" is produced. Could someone help me
understand why I get this result?

Because there are two fits involved, and one of them does not converge. 
  Because you have an intercept and offset in your model, the null fit 
also needs to be done, and that fit did not converge.

So the main results from your fit are reliable, but the null.deviance 
component is not.  Increase maxit to allow the null fit to converge.

Re: Optimization inconsistencies

Hans W Borchers — 2012-05-18T07:10:41

Use the equality constraint to reduce the dimension of the problem by one.
Then formulate the inequality constraints and apply, e.g., constrOptim().
You can immediately write down and use the gradient, so method "L-BFGS-B" is
appropriate.

Because the problem is linear, there is only one maximum and no dependency
on starting values. 

If you had supplied some data and code (which packages did you try, and how?),
a more concrete answer would have been possible. My own test example worked
out of the box.

Hans Werner

Re: Integration of two dimension function

Petr Savicky — 2012-05-18T06:51:42

Hi.

The function f() is a 0/1 function. So, the integral may be 
reformulated as the area of the set of points (x, y) satisfying

  0.2 <= x <= 0.8
  0   <= y <= 1
  x*y <= c

In other words, it is the univariate integral of the function

  g(x) = min(1, c/x)

over the range x \in [0.2, 0.8]. This is 
     
  G <- function(cc) {
    ifelse(cc <= 0.2, cc*log(4), # cc*(log(0.8) - log(0.2))
           ifelse(0.2 < cc & cc < 0.8, (cc - 0.2) + cc*(log(0.8) - log(cc)), 0.8 - 0.2))
  }

What i get is that G(cc) = 0.025 for cc = 0.025/log(4).

I hope there are not too may errors in this calculation.

Petr Savicky.

Proc AutoReg (SAS like Output in R)

anil — 2012-05-18T04:11:00

Hi,

I want to find out, how can i get a SAS like output for Proc Autoreg with
AR(1) and AR(2) terms.


Thanks...

--
View this message in context: http://r.789695.n4.nabble.com/Proc-AutoReg-SAS-like-Output-in-R-tp4630468.html
Sent from the R help mailing list archive at Nabble.com.

Covariance matrix in R with non-numeric variables

2012-05-18T04:33:46

Dear R help forum members,

I am modeling a gaussian distribution for a computational biology application and I am working in the statistical package "R". In this regard, my problem is that I have to construct a covariance matrix with variables (non-numeric) and the covariance matrix is to be used in an maximizers of the likelihood function to predict the variables in the matrix. I am unable to do that because I do not have an idea of how to construct a covariance matrix with non-numeric variable in the matrix.

Any help in this regard will be highly appreciated.

Thanks in advance.

Regards, B.Nataraj


[[alternative HTML version deleted]]

Re: Correlation Matrix

mahdi — 2012-05-17T20:37:27

thanks a lot dear. I will keep your advice in my mind.

--
View this message in context: http://r.789695.n4.nabble.com/Correlation-Matrix-tp4630389p4630448.html
Sent from the R help mailing list archive at Nabble.com.

Financial Statements Date Subsetting

Brian Edmundson — 2012-05-18T01:53:41

Dear All,

I'm new at R, but I really just need a couple of things.  The first thing I
need is to figure out how to get each individual financial statement
(CF,BS,IS).  I need each individual one because getting them all at once
allows for formatting issues once it is a CSV.  The date subsetting is what
I need because I will be running a statistical model in excel.  I know I
could probably build a cleaner model in R, but I have already built it in
excel.

The first place I am stuck at is actually viewing the financials. I have
gotten it to work a few times, but i am trying to script the installation
sequence ( I still assume I have to load the packages as well - which I
have done).

I have tried this sequence downloading straight from the - i think - the
CRAN mirror and I have also tried downloading this sequence straight from
R-Project.com...

install.packages("TTR")
install.packages("zoo")
install.packages("Defaults")
install.packages("xts")
install.packages("quantmod")

this still (from both downloading s

Correlation in Rattle

avideh yesharim — 2012-05-17T21:32:14

Hi,

I recently installed Rattle for R 2.15.0 and all the functions work properly except for the Correlation button. After I choose the Correlation radio button in Explore tab, I click Execute, but nothing happens. No results at all. What is it that I am missing?

Thank you,
Avideh

[[alternative HTML version deleted]]

Re: Failure building any package

Joshua Wiley — 2012-05-18T04:10:39

Hi Noah,

Can you put your package on github or at least upload the tar ball?
Although I agree with Steven that as is, the package will not pass
cran checks, that is not the error I would expect.

Some of the experts may have seen this before or know instantly, but
for everyone else, being able to try what you are trying may help.
Its also not a bad idea to have your code hosted somewhere anyway, so
you can slice two beets with one stroke as it were.

Cheers,

Josh

On Thu, May 17, 2012 at 7:44 PM, Noah Silverman <noahsilverman< at >ucla.edu> wrote: