similar to: factors in model.frame.default

# r-bugs@r-project.org `predict' complains about new factor levels, even if the "new" levels are merely levels in the original that didn't occur in the original fit and were sensibly dropped, and that don't occur in the prediction data either. (At least if `drop.unused.levels' was set to TRUE, which the default.) test> scrunge.data.2_ data.frame( y=runif( 3),

I am using an algorigm to split my data set into two random sections repeatedly and constuct a model using rpart() on one, test on the other and average out the results. One of my variables is a factor(crop) where each crop type has a code. Some crop types occur infrequently or singly. when the data set is randomly split, it may be that the first data set has a crop type which is not present in

Hi. I uses rpart to build a regression tree. Y is continuous. Now, I try to predict on a new set of data. In the new set of data, one of my x (call Incoterm, a factor) has a new level. I wonder why the error below appears as the guide says "For factor predictors, if an observation contains a level not used to grow the tree, it is left at the deepest possible node and

There is a problem in MacOS X with multiple definitions of the same symbol in different files that will be put into the same bundle or dynamic library by the dynamic linker. It occurs, for example, in the rpart package, which includes rpart.h in all its source files, and rpart.h has definitions of a structure rp and functions rp_init and so on. I think the same problem occurs in the grid package

Hi All, I'm a newbie and have two questions. Please pardon me if they are very basic. 1. I'm using a regression tree to predict the selling prices of 10 new records (homes). The following code is resulting in an error message: pred <- predict(model, newdata = outOfSample[, -6]) The error message is: Error in model.frame.default(Terms, newdata, na.action = na.action, xlev =

Philipp Benner reported a Debian bug report against r-cran-rpart aka rpart. In short, the issue has to do with how rpart evaluates a formula and supporting arguments, in particular 'weights'. A simple contrived example is ----------------------------------------------------------------------------- library(rpart) ## using data from help(rpart), set up simple example myformula <-

Hi, I have a dataframe, y, with 1000 rows and 266 columns. The first column of y is a 2 level factor response; columns 2:266 are each ordinal variables that can take on one of the values 1, 2, or 3. I'm trying to fit a model with the calls: (A) y.tree <- tree(pref~.,data=y,mindev=0,minsize=2) and (B) y.tree <- tree(pref~.,data=y,split='gini') pref is the name of the factor

Over a decade ago there was a problem with model.frame when the variable names were long: https://stat.ethz.ch/pipermail/r-help/2002-August/024492.html I have similar symptoms with R 2.15.3 on Windows 7: Browse[2]> x <- model.matrix(formula(myform), p$data) Error in model.frame.default(object, data, xlev = xlev) (from mice.R#601) : object is not a matrix My attempt at a work-around

Hi all, I''m still not quite happy with the NA and factor handling of lm and predict.lm in R1.6.1 (forcing me to use my not very skillfully crafted patches). Here is the problem 1: >

(I'm sending to r-bugs because rpart is one of the recommended packages and is always installed. I'm also sending it directly to Dr. Ripley, as the maintainer.) rpart working as a classifier does not work (produces no splits) when the class indicator has no instances of one of the factor levels, as long as the factor level is not the final level. I have at least a partial fix, which I

I just downloaded version 1.0.0 and several binary libraries (VR, rpart, norm, stataread) - WinNT version. I then converted a file from Stata 6.0 to R format by using the stataread library. The file converts perfectly and I was able to use the VR function lda on the dataframe without difficulty. I then tried to use the same dataframe with RPART. The model statement:

Dear r-devel list members, I'm posting this message here because it concerns the nlme package, which is maintained by R-core. The problem I'm about to describe is somewhere between a bug and a feature request, and so I thought it a good idea to ask here rather posting a bug report to the R bugzilla. I was made aware (by Ben Bolker) that the car::Anova() method for "lme"

Hi all, Does any of you know how to make a decision tree when the data set contains factors and numericals? I've got a data frame with 3 columns, where y and x1 are numerical and x2 contains factors. Is it possible to use the rpart package, and in that case how? Otherwise, is there another alternative? This is what I've tried so far > rpart(LT50_NA ~ Raf + Antho,

Dear list members, After further testing, I found that the following simplified version of model.matrix.lme(), which omits passing xlev to the default method, is more robust. The previous version generated spurious warnings in some circumstances. model.matrix.lme <- function(object, ...){ data <- object$data if (is.null(data)){ NextMethod(formula(object),

I think there are two bugs in aov() that shows up when the right hand side of `formula' contains both `-1' and an Error() term, e.g., aov(y ~ a + b - 1 + Error(c), ...). Without `-1' or `Error()' there is no problem. I've included and example, and the source of aov() with suggested fixes below. The first bug (labeled BUG 1 below) creates an extra, empty stratum inside

I believe you are right, but can you please explain why anyone would want to fit this model? It differs only in the coding from aov(y ~ a + b + Error(c), data=test.df) and merely lumps together the top two strata. There is a much simpler fix: in the line if(intercept) nmstrata <- c("(Intercept)", nmstrata) remove the condition (and drop the empty stratum later if you

Hi, I'm trying to get sparse.model.matrix to retain unused levels. I can't seem to get this working through the most obvious routes such as specifying drop.unused.levels = FALSE in the model.frame or trying to pass all levels in xlev,which is an argument to sparse.model.matrix (see code below). Any help would be gratefully received. Cheers, Jarrod fac<-factor(rep(1:10,10),

Seems to be a day for finding peculiar little things. There is an inconsistency in the behavior of lm vis a vis glm: > x <- rnorm(15) > y <- 1 + 10*x + rnorm(15) > z <- as.factor(rep(c("A","B","C"),rep(5,3))) > xyz <- data.frame(x,y,z) > fit.lm <- lm(y ~ x + z, data=xyz, subset=(z != "C")) > fit.glm <- glm(y ~ x +

Last Friday, I noticed that it is difficult to work with regression models in which there are factors. It is easier to do the old fashioned thing of coding up "dummy" variables with 0-1 values. The predict function's newdata argument is not suited to insertion of hypothetical values for the factor, whereas it has no trouble with numeric variables. For example, if one uses a

Dear DevelopeRs, in continuing with my suite of packages on experimental design, I am stuck with an issue that appears to be related to package AlgDesign - I have tried to get it solved by Bob Wheeler, but he seems to be stuck as well. Whenever AlgDesign is loaded, some of my code does not work any more. For example, in a fresh R session: require(DoE.base) fac.design(nlevels=c(2,6,2))