similar to: how to train ksvm with spectral kernel (kernlab) in caret?

I am used to packages like e1071 where you have a tune step and then pass your tunings to train. It seems with caret, tuning and training are both handled by train. I am using train and trainControl to find my hyper parameters like so: MyTrainControl=trainControl( method = "cv", number=5, returnResamp = "all", classProbs = TRUE ) rbfSVM <- train(label~., data =

I have never been able to get class probabilities to work and I am relatively new to using these tools, and I am looking for some insight as to what may be wrong. I am using caret with kernlab/ksvm. I will simplify my problem to a basic data set which produces the same problem. I have read the caret vignettes as well as documentation for ?train. I appreciate any direction you can give. I

kernlab version 0.9-7 is now online and incorporates : + a much improved fast implementation of string kernels "stringdot" based on suffix arrays. + a new kernel method kmmd() which implements a non-parametric kernel based two sample test. The new kernlab version also includes many minor improvements and fixes. Alexandros _______________________________________________

kernlab version 0.9-7 is now online and incorporates : + a much improved fast implementation of string kernels "stringdot" based on suffix arrays. + a new kernel method kmmd() which implements a non-parametric kernel based two sample test. The new kernlab version also includes many minor improvements and fixes. Alexandros _______________________________________________

Dear all, I am using kernlab package in R, and I have amino acid sequences with different lenghts as input for a SVM and I need to go through this sequences using windows (sliding or fixed) of size X. Does anyone has any suggestions about which function I should use? I thought I could use stringdot, but I am not sure whether it will do what I need.., I have defined my stringdot as: mystringdot

I have a binary classification problem where the fraction of positives is very low, e.g. 20 positives in 10,000 examples (0.2%) What is an appropriate cross validation scheme for training a classifier with very few positives? I currently have the following setup: ======================================== library(caret) tmp <- createDataPartition(Y, p = 9/10, times = 3, list = TRUE)

Hi!! I am trying to figure out how to use the string kernel "stringdot" in kernlab. k <- function(x,y) { (sum(x*y) +1)*exp(-0.001*sum((x-y)^2)) } class(k) <- "kernel" data(promotergene) ## train svm using custom kernel gene.k <- ksvm(Class~.,data=promotergene,kernel=k,C=10,cross=5) # works fine in this case gene.rbf <-

Dear developers, I have jutst started working with caret and all the nice features it offers. But I just encountered a problem: I am working with a dataset that include 4 predictor variables in Descr and a two-category outcome in Categ (codified as a factor). Everything was working fine I got the results, confussion matrix etc. BUT for obtaining the AUC and predicted probabilities I had to add

Hi, A. In a nutshell: The training error, obtained as "error (ret)", from the return value of a ksvm () call for a eps-svr model is (likely) being computed wrongly. "nu-svr" and "eps-bsvr" suffer from this as well. I am attaching three files: (1) ksvm.R from the the kernlab package, un-edited, (2) ksvm_eps-svr.txt: (for easier reading) containing only eps-svr

Hello list, I am faced with a two-class classification problem with highly asymetric class sizes (class one: 99%, class two: 1%). I'd like to obtain a class probability model, also introducing available information on the class prior. Calling kernlab/ksvm with the line > ksvm_model1<-ksvm(as.matrix(slides), as.factor(Class), class.weights= c("0" =99, "1" =1),

Hello, I would like to know, whether for the support vector classification function ksvm() the response values stored in object at ymatrix are cross validated outputs/predictions: Example code from package kernlab, function ksvm: library(kernlab) ## train a support vector machine filter <- ksvm(type~.,data=spam,kernel="rbfdot",kpar=list(sigma=0.05),C=5,cross=3) filter filter at

Dear list, I am using the ksvm function from kernlab as follows: (1) learning > svm.pol4 <- ksvm(class.labs ~ ., data = train.data, prob.model = T, scale = T, kernel = "polydot") (2) prediction > svm.pol.prd4 <- predict(svm.pol4, train.data, type = "probabilities")[,2] But unfortunately, when calling the prediction, once in every 10s of times (using the exact

Hello all, I'm trying to run a gird parameter search for a svm. Therefore I'M using the ksvm function from the kernlab package. ---- svp <- ksvm(Ktrain,ytrain,type="nu-svc",nu=C) ---- The problem is that the optimization algorithm does not return for certain parameters. I tried to use setTimeLimit() but that doesn't seem to help. I suspect that ksvm() calls c code that

Hi there, I'm trying to do a regression experiment on a multidimensional dataset where both x and y in the model are multidimensional vectors. I'm using R version 2.9.2, updated packages, on a Linux box. I've tried gausspr(), ksvm() and rvm(), and the models are computed fine, but I'm always getting the same error message when I try to use predict(): "Error in

Hello, together I'm trying to use user defined kernel. I know that kernlab offer user defined kernel(custom kernel functions) in R. I used data spam including package kernlab. (number of variables=58 number of examples =4061) i'm user defined kernel's form, kp=function(d,e){ as=v*d bs=v*e cs=as-bs cs=as.matrix(cs) exp(-(norm(cs,"F")^2)/2) }

When I try to crate a grid of parameters for training with caret I get various errors: ------------------------------------------------------------ > my_grid <- createGrid("rf") Error in if (p <= len) { : argument is of length zero > my_grid <- createGrid("rf", 4) Error in if (p <= len) { : argument is of length zero > my_grid <-

I have data set up like the following: control1 <- sample(1:75, 3947398, replace=TRUE) control2 <- sample(1:75, 28793, replace=TRUE) control3 <- sample(1:100, 392733, replace=TRUE) control4 <- sample(1:75, 858383, replace=TRUE) patient1 <- sample(1:100, 28048, replace=TRUE) patient2 <- sample(1:50, 80400, replace=TRUE) patient3 <- sample(1:100, 48239, replace=TRUE) control

I'm using caret to assess classifier performance (and it's great!). However, I've found that my results differ between R2.* and R3.* - reported accuracies are reduced dramatically. I suspect that a code change to kernlab ksvm may be responsible (see version 5.16-24 here: http://cran.r-project.org/web/packages/caret/news.html). I get very different results between caret_5.15-61 +

Hello all, I have searched and have not yet identified a solution so now I am sending this message. In short, I need to split my data into training, validation, and testing subsets that keep all observations from the same sites together ? preferably as part of a cross validation procedure. Now for the longer version. And I must confess that although my R skills are improving, they are not so

Dear all, I am using the "caret" Package for predictors selection with a randomForest model. The following is the train function: rfFit<- train(x=trainRatios, y=trainClass, method="rf", importance = TRUE, do.trace = 100, keep.inbag = TRUE, tuneGrid = grid, trControl=bootControl, scale = TRUE, metric = "ROC") I wanted to use ROC as the metric for variable