similar to: Bug in predict(newdata=x) with poly() (PR#1258)

I am fitting a regression model with a bs term and then making predictions based on the model. According to some info on the internet at http://www.stat.auckland.ac.nz/~yee/smartpred/DummiesGuide.txt there are some problems with using predict.lm when you have a model with terms such as bs,ns,or poly. However when I used one of the examples they said would illustrate the problems I get virtually

Hi I am having a particular problem with some glm models I am running. I have been adapting code from Bill Venables 'Programmers niche' in RNews Vol 2/2 to fit ca. 1000 glm models to a combination of species 0/1 data (as Y) and related physicochemical data (X), to automate the process of fitting this many models. I have successfully managed to fit all the models and have stored the

I have a classification tree model similar to the following (slightly simplified here): > treemod<-tree(y~x) where y is a factor and x is a matrix of numeric predictors. They have dimensions: > length(y) [1] 1163 > dim(x) [1] 1163 75 I?ve evaluated the tree model and am happy with the fit. I also have a matrix of cases that I want to use the tree model to classify. Call it

Thanks for the reproducable example. I can confirm that it fails on my machine using survival 2-37.5, the next soon-to-be-released version, The issue is with NextMethod, and my assumption that the called routine inherited everything from the parent, including the environment chain. A simple test this AM showed me that the assumption is false. It might have been true for Splus. Working this

Full_Name: Renaud Lancelot Version: Version 2.3.0 (2006-04-24) OS: MS Windows XP Pro SP2 Submission from: (NULL) (82.239.219.108) I think there is a bug in predict.lme, when a polynomial generated by poly() is used as an explanatory variable, and a new data.frame is used for predictions. I guess this is related to * not * using, for predictions, the coefs used in constructing the orthogonal

Here's my little discussion example for a quadratic regression: http://pj.freefaculty.org/R/WorkingExamples/regression-quadratic-1.R Students press me to know the benefits of poly() over the more obvious regression formulas. I think I understand the theory on why poly() should be more numerically stable, but I'm having trouble writing down an example that proves the benefit of this. I

I have a function that fits polynomial models for the orders in n: lmn <- function(d,n){ models=list() for(i in n){ models[[i]]=lm(y~poly(x,i),data=d) } return(models) } My data is: > d=data.frame(x=1:10,y=runif(10)) So first just do it for a cubic: > mmn = lmn(d,3) > predict(mmn[[3]]) 1 2 3 4 5 6 7 8

Hello R Gurus, I am perplexed by the different results I obtained when I ran code like this: set.seed(100) test1<-randomForest(BinaryY~., data=Xvars, trees=51, mtry=5, seed=200) predict(test1, newdata=cbind(NewBinaryY, NewXs), type="response") and this code: set.seed(100) test2<-randomForest(BinaryY~., data=Xvars, trees=51, mtry=5, seed=200, xtest=NewXs, ytest=NewBinarY) The

Full_Name: Murray H Smith Version: 1.6.1 OS: Windows Submission from: (NULL) (202.36.29.1) This is report is more of a matter of completeness rather than an outright bug. The predict function does not handle the prediction from the constant model appropriately. It also differs from Splus in this respect. The length of the vector (or first dimension of the matrix, if type = "terms" is

Estimado Oliver Nuñez Envío un ejemplo reproducible. Javier Marcuzzi # de donde tomo datos, y tiene el modelo (en el pdf) library(MCMCglmm) # librería con las funciónes que voy a usar library(nlme) datos0<-ChickWeight # creo algunos datos que agrego a los origonales Factor<-as.numeric(datos0$Chick) Factor[Factor > 0 & Factor <= 10] <- 'A' Factor[Factor > 10

Dear r-devel list members, I've recently encountered the following problem using predict() with a model that has raw-polynomial terms. (Actually, I encountered the problem using model.frame(), but the source of the error is the same.) The problem is technical and concerns the design of poly(), which is why I'm sending this message to r-devel rather than r-help. To illustrate:

Dear R experts, I am observing undesired behavior of predict(fit, newdata), in case when fit object is produced by lm() involving a poly(). Here is how to reproduce: x <- c(1:10) y <- sin(c(1:10)) fit <- lm(formula=y~poly(x, 5, raw=TRUE)) predict(fit, newdata=data.frame(x=c(1:10))) ## this works predict(fit, newdata=data.frame(x=c(1:1))) ## this is broken, error below Error in poly(x,

Colleagues, I am using "nls" successfully (2.11.1, OS X) but I am having difficulties retrieving part of the output - residual sum of squares. I have assigned the output to FIT: > > FIT > Nonlinear regression model > model: NEWY ~ PMESOR + PAMPLITUDE * cos(2 * pi * (NEWX - POFFSET)/PERIOD) > data: parent.frame() > PMESOR PAMPLITUDE POFFSET >

Dear All, When I plot the values and linear regression line for one data set, it is fine. But for another one I see zigzags, when I plot the confidence interval >cd Depth CHAOsep12RNA 9,94 804 25,06 1476,833333 40,04 1540,561404 50,11 1575,166667 52,46 349,222222 54,92 1941,5 57,29 1053,507042 60,11 1535,1 70,04 2244,963303 79,97 1954,507042 100,31 2679,140625 >

Hi, I am segfaulting when using predict.loess() (checked with r62092). I've traced the source with the help of valgrind (output pasted below) and it appears that this is due to int overflow when allocating an int work array in loess_workspace(): liv = 50 + ((int)pow((double)2, (double)D) + 4) * nvmax + 2 * N; where liv is an (global) int. For D=1 (one x variable), this overflows at

This smells like a bug to me. The error is triggered by the line: variables <- eval(predvars, data, env) inside model.frame.default(). At that point, na.action has not been applied, so poly() ended being called on data that still contains missing values. The qr() that issued the error is for generating the orthogonal basis when evaluating poly(), not for fitting the linear model itself.

This smells like a bug to me. The error is triggered by the line: variables <- eval(predvars, data, env) inside model.frame.default(). At that point, na.action has not been applied, so poly() ended being called on data that still contains missing values. The qr() that issued the error is for generating the orthogonal basis when evaluating poly(), not for fitting the linear model itself.

Hello, I've executed the following predict.coxph function to enable prediction for new variable values (error is included). *predict(cox_out,newdata=data.frame(Meter3.Value=100.001, Meter4.Value=200.001,Meter5.Value=300.001,Meter10.Value= 400.001,type="expected")) Error in model.frame.default(data = data.frame(Meter3.Value = 100.001, : variable lengths differ (found for

Hi, I am curious about how to interpret the results of a polynomial regression-- using poly(raw=TRUE) vs. poly(raw=FALSE). set.seed(123456) x <- rnorm(100) y <- jitter(1*x + 2*x^2 + 3*x^3 , 250) plot(y ~ x) l.poly <- lm(y ~ poly(x, 3)) l.poly.raw <- lm(y ~ poly(x, 3, raw=TRUE)) s <- seq(-3, 3, by=0.1) lines(s, predict(l.poly, data.frame(x=s)), col=1) lines(s,

Hello all, I have generated a principal components regression model using the pcr() function from the PLS package (R version 2.12.0). I am getting a "non-conformable arguments" error when I try to use the predict() function on new data, but only when I try to read in the new data from a separate file. More specifically, when my data looks like this #########training data