ays have a
record when you discover something you want to reproduce.
6/ If parallel computing is considered then for reproducibility you need
to save the number of nodes in the cluster. (I think this point is not
as widely known as it should be.)
No doubt I have forgotten a few things.
Paul Gil
On 03/04/2018 07:14 PM, Henrik Bengtsson wrote:
On Sun, Mar 4, 2018 at 3:23 PM, Duncan Murdoch wrote:
...
An issue is that .Random.seed doesn't contain the full state of the RNG
system, so restoring it doesn't necessarily lead to an identical sequence of
output. The only way to guarantee th
e it does not seem to be widely
appreciated: the number of nodes affects the random stream, so recording
the number of compute nodes along with the RNG and seed information is
important for reproducible results. This has the unfortunate consequence
that an experiment cannot be reproduced on a la
On 02/10/2018 06:00 AM, r-help-requ...@r-project.org wrote:
Did you check the gradient? I don't think so. It's zero, so of course
you end up where you start.
Try
data.input= data.frame(state1 = (1:500), state2 = (201:700) )
err.th.scalar <- function(threshold, data){
state1 <- data$stat
On 11/17/2016 04:49 PM, Jeff Newmiller wrote:
require(tfplot)
tfplot(x.ts)
Would just like to point out that require() should not be treated as
interchangeable with library(). The former returns a logical status
indicating success or failure, while the latter throws an error if it
falls. You
On 11/17/2016 06:00 AM, r-help-requ...@r-project.org wrote:
Hi,
As I sit and learn how to work with time series, I've run into a problem
that is eluding a quick easy answer (and googling for answers seems to
really slow the process...)
Question #1--
In a simple example on R 3.3.1 (sorry my emp
I am trying to deal with a maximisation problem in which it is possible
for the objective function to (quite legitimately) return the value
-Inf,
(Just to add to the pedantic part of the discuss by those of us that do
not qualify as younger and wiser:)
Setting log(0) to -Inf is often conveni
Narendra
You should check the definition of your objective function, something
seems to be wrong. When you evaluate the objective function with your
initial guess, and then with some different value, you should expect the
returned value will be different, but it is not:
> print(Koval.Error.
-%d %H:%M:%S") )
There are several other details in the function that you may find useful.
Paul Gilbert
Date: Mon, 5 Sep 2016 22:28:50 +0200
From: Stef Mientki
hello,
I've a number of timeseries into a database and want to display these
timeseries into graph.
Now the code below work
See also package dse. There are examples in the guide:
http://cran.at.r-project.org/web/packages/dse/vignettes/Guide.pdf
Paul
On 02/14/2016 06:00 AM, r-help-requ...@r-project.org wrote:
Date: Fri, 12 Feb 2016 18:12:37 + From: Thomas
Lofaro To:"r-help@R-project.org"
Subject: [R] FW: Multi
ckage "dlm"
Message-ID:
Content-Type: text/plain; charset="UTF-8"
Hi: I don't have time to look at the details of what you're doing but the
"equivalence"
between state space and arima ( as paul gilbert pointed out a few weeks ago
) is not a true equiva
Lorenzo
Berend's suggestion may be the simplest if you have univariate ts style
series. If you are writing code and you want it to work with
multivariate series and other time representations then you might want
to consider the splice() function in package tframe.
library(tfplot)
ts3 <- s
ot be trying to invert. In a
non-innovations form the state has a physical interpretation rather than
being an expectation, so the state tracking error should not be
degenerate in that case.
I mention all this because your covariance matrix looks very close to zero.
Paul Gilbert
On 12/11/2015 0
I was similarly bitten a couple of days ago when I submitted an update
of my deldir package. I'm*pretty* sure that I completed all the
requisite steps in the steps in the web-based submission procedure, but
I never received the promised "confirmation email". Being a thicko, I
overlooked this f
I'm not sure if this was due to mailer wrap, but I think you are putting
a between the end of the table name and the quote mark, and
possibly that or a space is being included in the table name in the
sqlQuery() statement. Do you have the same problem if you put the quote
mark immediately afte
A post with subject "Basic R to SQL considerations" has been put on
R-sig-DB to address some of the basic questions here. The R-sig-DB
thread is not specific to RPostgreSQL.
I have learned enough since this posting to be aware that my question was
based on assumptions so false to fact that th
I'm not sure if I understand exactly what you mean, but if you want
separate panels, one above the other, with a common time span but each
with their own Y scale, then the function tfplot() in package tfplot
does this. There are examples in the User Guide at
http://cran.at.r-project.org/web/pac
The responses to this seem to be assuming that you want users to have
access to your tt() function, that is, you export it. Just in case the
really simple case has been overlooked: if you are only using this
function internally in your package there should be no problem. Your
package's namespac
On 13-11-25 06:00 AM, r-help-requ...@r-project.org wrote:
Date: Sun, 24 Nov 2013 13:04:43 -0600
From: Yihui Xie
To: Bert Gunter
Cc:"r-help@r-project.org"
Subject: Re: [R] Should there be an R-beginners list?
Message-ID:
Content-Type: text/plain; charset=ISO-8859-1
I'm not aware of a
On 13-10-22 06:00 AM, Weiwu Zhang wrote:
My data is sampled once per minute.
At the same second each minute or not? Regularly spaced would mean
exactly one minute between observations.
There are invalid samples, leaving
a lot of holes in the samples, successful sample is around 80% of al
(from R-help Digest, Vol 117, Issue 26)
Date: Sun, 25 Nov 2012 07:49:05 -0800 (PST)
From: billycorg
To:r-help@r-project.org
Subject: [R] creation of an high frequency series
Message-ID:<1353858545067-4650744.p...@n4.nabble.com>
Content-Type: text/plain; charset=us-ascii
Hi R Users!
I would li
Matthieu
(regarding the final question, but not the ones before that.)
The short answer is that you seem to have a solution, since hessian in
numDeriv works. The long answer follows.
Outside a constraint area one cannot assume that a function will
evaluate properly. Some constraints are impo
On 12-10-15 06:00 AM, r-help-requ...@r-project.org wrote:
Jeff,
My understanding is that the lag command will lag an entire time series. That
isn't what I'm looking for.
I just want, for example, today, and 5 entries back.
for exmple:
iter <- '2011-05-18'
observations[iter] # works fin
Package dse also does vector auto-regression with exogenous variables
(and also vector ARMA and state-space models with exogenous variables).
But I don't understand what you mean by "not taking the base in the
model" so I don't know if it will solve your problem.
Paul
On 12-10-04 06:00 AM, r
On 12-09-17 06:00 AM, r-help-requ...@r-project.org wrote:
Date: Sun, 16 Sep 2012 14:41:42 -0500
From: Dirk Eddelbuettel
To: "Eberle, Anthony"
Cc:r-help@r-project.org
Subject: Re: [R] Question about R performance on UNIX/LINUX with
different memory/swap configurations
Message-ID:<2
On 12-01-14 06:00 AM, r-help-requ...@r-project.org wrote:
Date: Fri, 13 Jan 2012 02:09:09 -0800 (PST)
From: statquant2
To:r-help@r-project.org
Subject: Re: [R] simulating stable VAR process
Message-ID:<1326449349804-4291835.p...@n4.nabble.com>
Content-Type: text/plain; charset=us-ascii
Hello Pau
The situation for this kind of interface is much more advanced (for
economic time series data) than has been suggested in other postings.
Several of the organizations you mention support SDMX and I believe
there is a working R interface to SDMX which has not yet been made
public. A more complet
The simulate function in dse lets you specify the model and the
distribution of the noise term (or even their values so you can get any
distribution you like). So, you should be able to do what you want,
with either a VAR(p) or a vector ARMA process. If you are getting a
process that explodes
Tanja
Your TSdata object dfdata as printed in your email looks a bit funny. It should
be a list with a matrix of numeric data in the element named output:
> dfdata <- TSdata(output=matrix(rnorm(200), 100,2))
> dfdata
output data:
Series 1 Series 2
10.01905979 -0.096441240
20.
It seems more likely that the return value from your function is NA or NaN or
Inf. This might then result in an NA parameter value being calculated for the
next step. This is possible, for example, because the line search extends
outside the feasible region. You can re-write your function to ch
It is nearly impossible to go back after the fact and figure out the
seed you started with. So, you need to be careful to record the seed
first. If you are doing your simulations with a function then it is a
good idea to always start in the function by saving a record of the seed
and returning it w
Ted
I'm not sure I fully understand the question, but you may want to consider
creating a temporary table with a join, which you can do with a query from your
R session, and then query that table to bring the data into R. Roughly, the
logic is to leave the data in the db if you are not doing an
.
Paul Gilbert
>-Original Message-
>From: r-help-boun...@r-project.org [mailto:r-help-boun...@r-project.org]
>On Behalf Of Paul Miller
>Sent: April 8, 2010 11:14 AM
>To: r-help@r-project.org
>Subject: [R] Reading dates in R using SQL and otherwise (and
>someinteresting
dse2.
I have also taken this opportunity to do a long overdue cleanup of the Users'
Guides distributed as vignettes in the packages.
Paul Gilbert
La version française suit le texte an
I think the problem here is that you seem to be trying to specify a
non-innovations form model, for which both Q and R need to be specified,
but you have only specified Q.
The code guesses whether you are specifying an innovations model or an
non-innovations model based on whether you specify
Tom Schenk Jr wrote:
I use RODBC as my conduit from R to SQL. It works well when the tables are
stored on one channel, e.g.,
channel <- odbcConnect("data_base_01", uid="", dsn="")
However, I often need to match tables across multiple databases, e.g.,
"data_base_01" and "data_base_02". H
which is NFS mounted), or some other fix for this "feature"?
Paul Gilbert
La version française suit le texte anglais.
--
Gabor Grothendieck wrote:
Try
if (.Platform$OS.type == "windows") ... else ...
Gabor has suggested what I think is the best way to do this check, but
in my experience, if you are doing this check then you are almost
certainly missing some feature of R that will let you avoid doing it.
(Slightly off topic and outside my area - but this may be of interest to
the R community)
WASHINGTON (January 16, 2009) -- Today, a new international competition
called the "Digging into Data Challenge" was announced by four leading
research agencies: the Joint Information Systems Committee (J
Faheem Mitha wrote:
Hi Hans,
Thanks for the reply.
On Tue, 18 Nov 2008, Hans Werner Borchers wrote:
Faheem Mitha email.unc.edu> writes:
Hi,
Does anyone know of an R ORM (Object Relational Mapper)? I'm thinking of
something similar to sqlalchemy (http://www.sqlalchemy.org/).
Alternative
thanks to the authors and maintainers of the underlying packages,
and several people who have commented on early versions of the TS* packages.
Paul Gilbert
La version française suit le texte anglais
There will be a meeting of the Ottawa Gatineau R Users Group on Monday
November 10 from 4pm to 6pm. More details are provided below. Please
pass this message along to friends you think may be interested.
Paul Gilbert
_
Dear R friends:
To get OGRUG going again I've reserved roo
Christiane Reuter wrote:
Hi everybody,
I'm having a problem with connecting to my MySQL database.
Each time I try to connect
library(RMySQL)
m <- dbDriver("MySQL")
con <- dbConnect (m, host="my_host",username="my_username",
password="my_password", dbname="name_of_db")
it says
Fehler in mys
that to analyse ... the usual problem is getting
the data into R.
Paul Gilbert
Thanks,
rcoder
Ling, Gary (Electronic Trading) wrote:
Hi, here is one possible solution ...
-gary
### example ###
# create a 500x3 multi-ts
A <- matrix(rnorm(1500),nrow=500)
# misc. graphical setting
par(m
] $(FLAGS)/$@
@touch $(FLAGS)/$@ # better time stamp in some parallel situations
In this example I am assuming that printing z has zero length output if
everything was ok, but you could set up different situations, and you
will probably want to use try() in doSomething().
HTH.
Paul Gilbert
O
matrix is symmetric.
Paul Gilbert
Daomeng Gao wrote:
Hello R-help,
I need to compute matrices of first derivatives of a covariance matrix C
with entries given by c_ij=theta*exp(-0.5* sum(eta*(x[i,]-x[j,])^2)), wrt to
elements of eta, a m-dimensional vector of parameters, given a n*m data
matrix x
Davood Tofighi wrote:
> Thanks for your reply. For each condition, I will have a matrix or data
> frames of 1000 rows and 4 columns. I also have a total of 64 conditions for
> now. So, in total, I will have 64 matrices or data frames of 1000 rows and 4
> columns. The format of data I would like to
The system is also interesting in
the way that it has resolved one of the big problems of Statlib: there
is an automatic mechanism for removing broken and unmaintained packages.
Paul Gilbert
Kathy Gerber wrote:
> Earlier today I sent a question to Frank Harrell as an R developer with
> w
aces.
Another point that needs more emphasis is that R is actually a
programming language, like Matlab and and APL, so it really has more
general usefulness than "statistics packages" that one might use in the
narrower context of a st
a few example of
where you find this necessary, and ask if there is an OS independent way
to do it.
Paul Gilbert
> Would I be better off just to build two separate packages, please?
> If just one is needed, how could I determine which system is running
> in order to use the correct comma
appreciate comments on the relative merits from anyone that has more
experience in this area.
Paul Gilbert
Walter Paczkowski wrote:
> Good morning,
>
> I use SAS and R/S-Plus as my primary tools so I have a lot of experience with
> these programs. By far and away, SAS is superior for
me thought-provoking opinions on R being
> no-good and that you should write everything in C
People used to say assembler, that's progress.
Paul Gilbert
instead (mainly
> because R is slow and too good at graphics, encouraging data
> snooping). See http://fluff.info/blog/arch/
Taka
You might want to try R-SIG-Debian for this kind of question. Here are
instructions from Dirk in response to my similar question a few days ago.
HTH,
Paul
_
On 12 December 2007 at 18:17, Paul Gilbert wrote:
| I'm trying to build R from source on Ubuntu Gutsy Gibbon.
Giovanni Petris wrote:
> You may want to check package dlm and, possibly, dse.
>
Yes, you can also do this in dse, either in the ARMA specification or as
an equivalent state-space model. There is an example in the Users' Guide
distributed with the package.
Paul
> In dlm you can cast a VARMA
Mark W Kimpel wrote:
> I have identical R.profiles and R_HOME directories set up on both my
> local machine and a remote linux cluster. To keep my libraries and R
> install separate, I use a site-library on both machines.
>
> The first line of my .Rprofile is:
> '.libPaths(new= "~/R_HOME/site-
Giovanni Petris wrote:
> Kalman filter for general state space models, especially in its naive
> version, is known for its numerical instability. This is the reason
> why people developed square root filters, based on Cholesky
> decomposition of variance matrices. In package dlm the implementati
unless you have a very good starting
point, or are luck. This is a feature of the parameter space, not a
reflection on the optimization routine. One solution is to start by
estimating a VAR or ARMA model and then convert it to a state space
model, which was one of the original purposes of ds
Package dse does do linear, multivariate time series with multivariate
exogenous inputs, using VAR, multivariate ARMA, and state-space models,
just like you are describing. You can specify the model or have it
automatically determined. Perhaps you could give a few more details
about what you ar
(This could be fixed, it has not happened to me in a long time, but I
will mention it mainly because it is not something you are likely to
think of.)
It used to be that the X colours might be defined by the first
application that needed them, so if the systems administrator happened
to start u
59 matches
Mail list logo
