similar to: OT: Journal of Statistical Software

Hello list, I want to apply svm from library e1071, and I want to supply class weights. I do not really understand the help entry (and there is no example) class.weights: a named vector of weights for the different classes, used for asymetric class sizes. Not all factor levels have to be supplied (default weight: 1). All components have to be named. I have two

Dear list, on p. 11 of "Writing R-extensions" I read "See the ``Guidelines for Rd-files'' for guidelines ... which should be useful..." Where can I find these guidelines? Best, Christian -- *********************************************************************** Christian Hennig Seminar fuer Statistik, ETH-Zentrum (LEO), CH-8092 Zuerich (current) and Fachbereich

Dear list, what is the fastest way to compute a multivariate mean and cov-matrix? I presume that the mean is computed in cov, so it may be a waste of time to compute the mean first and then a second time inside of cov. Is it faster to use cov.wt, which gives cov-matrix and center? And: If mean and cov should be computed on a part of the data, is it faster to use cov.wt with some weights zero, or

Dear list, here is something I do not understand about cov.rob. > dat <- rmvnorm(200,rep(0,10),diag(10)) > cov.rob(dat,method="mcd") > cov.rob(dat,method="mcd",quantile.used= floor(3*211/4)) # All fine; default for quantile.used is floor(211/2) > dat <- rmvnorm(20,rep(0,10),diag(10)) > cov.rob(dat,method="mcd") # quantile.used is floor(31/2).

Hi, this is a second try to post this to the R-help mailing list. The first one has been rejected because of a too large attachment. Now I ask this without attaching the data. If you want to reproduce the results, please contact me directly to get the data. (First mail, rejected:) > Attached there is a 149*149 dissimilarity matrix; it is a file obtained by >

Dear group, I would like to document more than one function on one help page using the Rd language. I tried something like \value{ \code{foo1} returns a list with components \item{arg1}{Argument 1} \item{arg2}{Argument 2} \code{foo2} returns a list with components \item{arg3}{Argument 3} \item{arg4}{Argument 4} } Unfortunately this makes the text "\code{foo2} returns a list

Hi, I want to fit a kernel density estimator by bkde of library KernSmooth. I need only the density value at the point 0. I do not understand the following behaviour: > q <- rnorm(100) > bkq <- bkde(q, bandwidth=0.11, gridsize=1, range.x=c(0,0)) Error in 0:L : NA/NaN argument > bkq <- bkde(q, bandwidth=0.11, gridsize=1, range.x=c(-1,1)) > bkq $x [1] -1 $y [1] NA > bkq

Dear group, I want to use the mclust package on large data, and therefore I want to use a subset in the initial clustering phase. From help(mclust): k: If `k' is specified, the hierarchical clustering phase will use a sample of size `k' of the data in the initial hierarchical clustering phase. The default is to use the entire data set. m2 is a

Hi, I have run rpart to construct a regression tree. Is there any simple method to draw a nice picture of it, as it is usually done in books and paper to visualize the tree? Thank you, Christian *********************************************************************** Christian Hennig Fachbereich Mathematik-SPST/ZMS, Universitaet Hamburg hennig at math.uni-hamburg.de,

Hi, this is to announce a new version (1.1-2) of my package fpc. Apart from the stuff already present in the older version (methods for fixed point clustering and clusterwise regression, somewhat bug-cleaned and with faster examples) there is now a function cluster.stats, which computes some distance-based statistics often used for cluster validation, description and decision about the number of

Hi, this is to announce a new version (1.1-2) of my package fpc. Apart from the stuff already present in the older version (methods for fixed point clustering and clusterwise regression, somewhat bug-cleaned and with faster examples) there is now a function cluster.stats, which computes some distance-based statistics often used for cluster validation, description and decision about the number of

Hi, is there any method for goodness of fit testing of an (as general as possible) univariate distribution with parameters estimated, for normal, exponential, gamma distributions, say (e.g. the corrected p-values for the Kolmogorov-Smirnov or Chi-squared with corresponding ML estimation method)? It seems that neither ks.test nor chisq.test handle estimated parameters. I am aware of function

Hi, I would like to use the format function to get numbers all with three digits to the right of the decimal point, even in cases where there is no significant digit left. For example, I would like to get c(0.3456789,0.0000053) as "0.346" "0.000". It seems that it is not possible to force format to print a "0.000", i.e. without any significant decimal places. Is it

Hi, here are three results of random forest (version 4.0-1). The results seem to be more or less the same which is strange because I changed the classwt. I hoped that for example classwt=c(0.45,0.1,0.45) would result in fewer cases classified as class 2. Did I understand something wrong? Christian x1rf <- randomForest(x=as.data.frame(mfilters[cvtrain,]),

Hello list, I try to optimize some R code and it turns out that the function as.matrix takes a lot of time in my code. There is only one reason why I need as.matrix: My code should work for matrices with ncol=p and p should be allowed to be 1 or larger. Now I have a matrix x with dim(x)=(n,p), and I need to work with the submatrix y <-x[gv,], (gv a vector of n logicals) and to calculate

Hi, is there a test for the H0 skewness=0 (or with skewness as test statistic and normality as H0) implemented in R? Thank you, Christian *********************************************************************** Christian Hennig Fachbereich Mathematik-SPST/ZMS, Universitaet Hamburg hennig at math.uni-hamburg.de, http://www.math.uni-hamburg.de/home/hennig/

Dear list, dear Frank, I try to fit a Weibull survival regression model with package Design: sclear <- psm(sobj~V1+V2,dist="weibull") sobj is a one-dimensional survival object (no event indicators), V1 and V2 are factors. I get the following result: Parametric Survival Model: Weibull Distribution psm(formula = sobj ~ V1 + V2, dist = "weibull") Obs Events

Hi, the following strange error appears when I use qda: > qda1 <- qda(as.data.frame(mfilters[cvtrain,]),as.factor(traingroups)) Error: function is not a closure That's also strange: > qda1 <- qda(mfilters[cvtrain,],as.factor(traingroups)) Error in qda.default(mfilters[cvtrain, ], as.factor(traingroups)) : length of dimnames must match that of dims Some backgroud: >

Hi list, not really a problem to solve but a strange unwanted behavior (at least I do not like it): I'm loading library(multiv) to use sammon. Later I load library(MASS) which tells me that sammon is masked now, because there is also a MASS version of sammon. Since I believe that the MASS version is at least as good as the multiv version, I would also use the MASS version, but help(sammon)

Dear list, (also sent to Roger Bivand, but perhaps somebody of you can help me also) I am trying to use package spdep for fitting an SAR model with errorsarlm. However, I am not sure how to make a valid nb object out of my neighborhood. As far as I have seen, there is no documentation for nb.object. I have done the following: class(pschmid$nb) <- "nb" # pschmid is a prab object as