[R] PRESS and P2 statistics in R
Hello all, Is there any function in R by which I can calculate PRESS and P2 statistics for linear regression in R? Thanks Alex [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Re: [R] Caret package and lasso
Dear Max, Thanks for the reply. I will wait for your further comment on this. Regards Linda Garcia On Wed, Apr 7, 2010 at 8:03 PM, Max Kuhn mxk...@gmail.com wrote: Linda, Thanks for the example. I did this to make it more reproducible: set.seed(1) X-matrix(rnorm(50*100),nrow=50) y-rnorm(50*1) dimnames(X) colnames(X) - paste(V, 1:nrow(X)) # Applying caret package set.seed(2) con-trainControl(method=cv,number=10) data-NULL data- train(X,y, lasso, metric=RMSE,tuneLength = 10, trControl = con) I see your point here, but this code gives the same results: fit2 - enet(X, y, lambda = 0) predict(fit2, mode = fraction, s = data$bestTune$.fraction, type = coefficient)$coef (at least train() names the predictors). To me, it looks like enet is doing some filtering: dim(X) [1] 50 100 length(fit2$meanx) [1] 56 This appears to be independent of caret. I would contact the package maintainer off-list and ask. Max __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.htmlhttp://www.r-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code. [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] Biclustering package
Hello, I am looking for R package which can perform biclustering a part from biclust package. thanks Alex [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Re: [R] Biclustering package
Thank you very much Gabor. Alex On Thu, Feb 11, 2010 at 10:55 AM, Gábor Csárdi csa...@rmki.kfki.hu wrote: Alex, the isa2 package implements the biclustering algorithm discussed in Bergmann S, Ihmels J, and Barkai N. Iterative signature algorithm for the analysis of large-scale gene expression data. Phys Rev E Stat Nonlin Soft Matter Phys 2003 Mar; 67(3 Pt 1) 031902 Best, Gabor On Thu, Feb 11, 2010 at 10:51 AM, Alex Roy alexroy2...@gmail.com wrote: Hello, I am looking for R package which can perform biclustering a part from biclust package. thanks Alex [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.htmlhttp://www.r-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code. -- Gabor Csardi gabor.csa...@unil.ch UNIL DGM [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] glmnet in caret packge
Dear all, I want to train my model with LASSO using caret package (glmnet). So, in glmnet, there are two parameters, alpha and lambda. How can I fix my alpha=1 to get a lasso model? con-trainControl(method=cv,number=10) model - train(X, y, glmnet, metric=RMSE,tuneLength = 10, trControl = con) Thanks Alex Roy [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] How can I store the results
Dear R users, I am running a R code which gives me 10 columns and 160 rows. I need to run the code for 100 times and each time I need to store the results in a single file. I do not know how can I store them in a single file without over writting the results? Thanks Alex [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] penalization regression
Dear all, I am using penalization regression method for my data. I am wandering the following names are synonymous or not? Complexity parameter Penalty parameter Shrinkage factor Shrinkage parameter hyper parameter Thanks Alex [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] Elastic net in R (enet package)
Dear R users, I am using enet package in R for applying elastic net method. In elastic net, two penalities are applied one is lambda1 for LASSO and lambda2 for ridge ( zou, 2005) penalty. But while running the analysis, I realised tht, I optimised only one lambda. ( even when I looked at the example in R, they used only one penality) So, I am wandering which penalty they are referring to? Is it a combination of penalties or one of them. I read the paper of zou and hastie but still in doubt. Thanks in advance Alex [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] Permutation test and R2 problem
Hi, I have optimized the shrinkage parameter (GCV)for ridge and got my r2 value is 70% . to check the sensitivity of the result, I did permutation test. I permuted the response vector and run for 1000 times and draw a distribution. But now, I get r2 values highest 98% and some of them more than 70 %. Is it expected from such type of test? *I was under impression that, r2 with real data set will always maximum! And permutation will not be effected i.e. permuted r2 will always less than real one! * ** thanks a lot Alex [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] Rank of matrix
Dear all, Rank of a matrix depends on which factors? Only on rows or coumns? or both ? If there is a collinearlity in the variables ( columns ) does it effects the rank? X-matrix((rnorm(1)),50) dim(X) [1] 50 200 qr(X)$rank [1] 50 X[,2]-X[,30] qr(X)$rank [1] 50 X[10,]-X[7,] qr(X)$rank [1] 49 Thanks Alex [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] Collinearity in Linear Multiple Regression
Dear all, How can I test for collinearity in the predictor data set for multiple linear regression. Thanks Alex [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] Linear Regression Problem
Dear All, I have a matrix say, X ( 100 X 40,000) and a vector say, y (100 X 1) . I want to perform linear regression. I have scaled X matrix by using scale () to get mean zero and s.d 1 . But still I get very high values of regression coefficients. If I scale X matrix, then the regression coefficients will bahave as a correlation coefficient and they should not be more than 1. Am I right? I do not whats going wrong. Thanks for your help. Alex *Code:* UniBeta - sapply(1:dim(X)[2], function(k) + summary(lm(y~X[,k]))$coefficients[2,1]) pval - sapply(1:dim(X)[2], function(l) + summary(lm(y~X[,l]))$coefficients[2,4]) [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Re: [R] Linear Regression Problem
Dear Vito, Thanks for your comments. But I want to do Simple linear regression not Multiple Linear regression. Multiple Linear regression is not possible here as number of variables are much more than samples.( X is ill condioned, inverse of X^TX does not exist! ) I just want to take one predictor variable and regress on y and store regression coefficients, p values and R^2 values. And the loop go up to 40,000 predictors. Alex On Tue, Jul 14, 2009 at 5:18 PM, Vito Muggeo (UniPa) vito.mug...@unipa.itwrote: dear Alex, I think your problem with a large number of predictors and a relatively small number of subjects may be faced via some regularization approach (ridge or lasso regression..) hope this helps you, vito Alex Roy ha scritto: Dear All, I have a matrix say, X ( 100 X 40,000) and a vector say, y (100 X 1) . I want to perform linear regression. I have scaled X matrix by using scale () to get mean zero and s.d 1 . But still I get very high values of regression coefficients. If I scale X matrix, then the regression coefficients will bahave as a correlation coefficient and they should not be more than 1. Am I right? I do not whats going wrong. Thanks for your help. Alex *Code:* UniBeta - sapply(1:dim(X)[2], function(k) + summary(lm(y~X[,k]))$coefficients[2,1]) pval - sapply(1:dim(X)[2], function(l) + summary(lm(y~X[,l]))$coefficients[2,4]) [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.htmlhttp://www.r-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code. -- Vito M.R. Muggeo Dip.to Sc Statist e Matem `Vianelli' Università di Palermo viale delle Scienze, edificio 13 90128 Palermo - ITALY tel: 091 6626240 fax: 091 485726/485612 http://dssm.unipa.it/vmuggeo [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Re: [R] Linear Regression Problem
Dear Dr. Ravi Varadhan, Thanks for your comments. Here, variables (p) are in columns and samples are in rows(n). And I want to find out significant variables associated with response (y). The reason why I said multiple linear regression (MLR) is not possible : MLR or classical MLR developed with a philosophy where nr. of samples (n) are more than variables(p). And predictors are uncorrelated. So, I do not consider penalization/shrinkage/regularaization methods as a traditional regression methods such as MLR. The solutions you suggested, I am completely agree with it , even I can add some other techniques like Elastic net, Partial Least squares, Principal component regression etc or may be machine learning methods like Support vector regression or Random forest regression to get done my job. But I want to do Univarite Method ( like Simple Linear regression ) for some purpose. Regards Alex On Tue, Jul 14, 2009 at 5:48 PM, Ravi Varadhan rvarad...@jhmi.edu wrote: I am not sure that you really want to do separate regressions for each row of X, with the same y. This does not make much sense. Why do you think multiple linear regression is not possible just because X'X is not invertible? You have 2 main options here: 1. Obtain a minimum-norm solution using SVD (also known as Moore-Penrose inverse). This solution minimizes ||y - Xb|| subject to minimum ||b|| 2. Obtain a regularized solution such as the ridge-regression, as Vito suggested. You can do (1) as follows: require(MASS) soln - ginv(X, y) Here is an example: X - matrix(rnorm(1000), 10, 100) # matrix with rank = 10 b - rep(1, 100) y - crossprod(t(X), b) soln - c(ginv(X) %*% y) # this will not be close to b Hope this helps, Ravi. --- Ravi Varadhan, Ph.D. Assistant Professor, The Center on Aging and Health Division of Geriatric Medicine and Gerontology Johns Hopkins University Ph: (410) 502-2619 Fax: (410) 614-9625 Email: rvarad...@jhmi.edu Webpage: http://www.jhsph.edu/agingandhealth/People/Faculty_personal_pages/Varadhan.h tml -Original Message- From: r-help-boun...@r-project.org [mailto:r-help-boun...@r-project.org] On Behalf Of Alex Roy Sent: Tuesday, July 14, 2009 11:29 AM To: Vito Muggeo (UniPa) Cc: r-help@r-project.org Subject: Re: [R] Linear Regression Problem Dear Vito, Thanks for your comments. But I want to do Simple linear regression not Multiple Linear regression. Multiple Linear regression is not possible here as number of variables are much more than samples.( X is ill condioned, inverse of X^TX does not exist! ) I just want to take one predictor variable and regress on y and store regression coefficients, p values and R^2 values. And the loop go up to 40,000 predictors. Alex On Tue, Jul 14, 2009 at 5:18 PM, Vito Muggeo (UniPa) vito.mug...@unipa.itwrote: dear Alex, I think your problem with a large number of predictors and a relatively small number of subjects may be faced via some regularization approach (ridge or lasso regression..) hope this helps you, vito Alex Roy ha scritto: Dear All, I have a matrix say, X ( 100 X 40,000) and a vector say, y (100 X 1) . I want to perform linear regression. I have scaled X matrix by using scale () to get mean zero and s.d 1 . But still I get very high values of regression coefficients. If I scale X matrix, then the regression coefficients will bahave as a correlation coefficient and they should not be more than 1. Am I right? I do not whats going wrong. Thanks for your help. Alex *Code:* UniBeta - sapply(1:dim(X)[2], function(k) + summary(lm(y~X[,k]))$coefficients[2,1]) pval - sapply(1:dim(X)[2], function(l) + summary(lm(y~X[,l]))$coefficients[2,4]) [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.htmlhttp://www.r-project.org/posting-guide.html http://www.r-project.org/posting -guide.html and provide commented, minimal, self-contained, reproducible code. -- Vito M.R. Muggeo Dip.to Sc Statist e Matem `Vianelli' Universit` di Palermo viale delle Scienze, edificio 13 90128 Palermo - ITALY tel: 091 6626240 fax: 091 485726/485612 http://dssm.unipa.it/vmuggeo [[alternative HTML version deleted]] [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting
Re: [R] Random Forest Variable Importance Interpretation
Hi, Are you looking for variable selection? If this is the case than you can use LASSO, Elastic net, Sparse PLS regression methods which encourages variable selection. PCA does not select variables as you get all your variables in the PCs. You can sparse PCA. Regards Alex On Wed, Jun 24, 2009 at 6:04 PM, lara harrup (IAH-P) lara.har...@bbsrc.ac.uk wrote: Hi I am trying to explore the use of random forests for regression to identify the important environmental/microclimate variables involved in predicting the abundance of a species in different habitats, there are approx 40 variable and between 200 and 500 data points depending on the dataset. I have successfully used the randomForest package to conduct the analysis and looked at the %IncMSE and IncNodeImpurity values given by calling and plotting these out and have looked at the partial dependence plots for the different variables effect of the response but I have been looking though the literature to see how people have previously used this type of analysis and I would like to be able to plot out the overall variable importance in some form of PCA Scree graph but havn't got a clue how to even start this so any suggestions will be most appreciated? Many thanks in advance Lara [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.htmlhttp://www.r-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code. [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] How to save multiple images??
Dear all, How can I save multiple images in my working directory?? I used save.image() but could not succeeded. Thanks in advance Alex [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] How to write loop
Dear all, I want to do the following process as a loop ( to run automatically with dimension of X, here 50). How can I do that? Your cooments will be highly appreciable. Alex *# Code:* library(lars) library(chemometrics) X-matrix(rnorm(2500),ncol=50) dim(X) # [1] 50 50 X1-X[,2:dim(X)[2]] # I have taken out first column dim(X1) #[1] 50 49 X2-X1[2:dim(X1)[1],] # new X2 is constructed dim(X2) #[1] 49 49 y-as.matrix(X1[1,]) # Now first row of the X1 acts a response vector dim(y) # [1] 49 1 # application of LASSO regression where y is response and X2 is a design matrix data1-data.frame(y,X2=I(X2)) lasso_res=lassoCV(y~X2,data=data1,K=10,fraction=seq(0.1,1,by=0.1),use.Gram=FALSE) # to get optimum value of Cross Validation lasso_coef=lassocoef(y~X2,data=data1,sopt=lasso_res$sopt,use.Gram=FALSE) # to get the coefficients [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Re: [R] How to write loop
Hi Jiim, Thanks . I want to do the following: 1. each time I need to drop one column, say first column 1 from matrix X. 2 then take out row 1 of the remainning matrix and that row becomes response (y) 3. do lasso regression on remaining X to y. 4. store the coefficients Similarly, in next run 1. I need to drop 2nd column, from matrix X. 2 then take out row 2 of the remainning matrix and that row becomes response (y) 3. do lasso regression on remaining X ( in example: X2to y.) 4. store the coefficients repeat On Sat, Jun 13, 2009 at 7:19 PM, jim holtman jholt...@gmail.com wrote: It is not exactly clear what you want to iterate on. What is going to be changing each time through the loop? On Sat, Jun 13, 2009 at 10:09 AM, Alex Roy alexroy2...@gmail.comwrote: Dear all, I want to do the following process as a loop ( to run automatically with dimension of X, here 50). How can I do that? Your cooments will be highly appreciable. Alex *# Code:* library(lars) library(chemometrics) X-matrix(rnorm(2500),ncol=50) dim(X) # [1] 50 50 X1-X[,2:dim(X)[2]] # I have taken out first column dim(X1) #[1] 50 49 X2-X1[2:dim(X1)[1],] # new X2 is constructed dim(X2) #[1] 49 49 y-as.matrix(X1[1,]) # Now first row of the X1 acts a response vector dim(y) # [1] 49 1 # application of LASSO regression where y is response and X2 is a design matrix data1-data.frame(y,X2=I(X2)) lasso_res=lassoCV(y~X2,data=data1,K=10,fraction=seq(0.1,1,by=0.1),use.Gram=FALSE) # to get optimum value of Cross Validation lasso_coef=lassocoef(y~X2,data=data1,sopt=lasso_res$sopt,use.Gram=FALSE) # to get the coefficients [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.htmlhttp://www.r-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code. -- Jim Holtman Cincinnati, OH +1 513 646 9390 What is the problem that you are trying to solve? [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] Heatmap
Hello Group, How can I draw heatmap with variable names in the plot? Thanks Alex [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] Caret package: coeffcients for regression
Dear All, I am using Caretpackage for SVM regression and elastic net regression . I can get the final fiited vs observed values. How can I get the coefficients? Any ideas? Thanks Alex [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] loop problem for extract coefficients
Dear R users, I have problem with extracting coefficients from a object. Here, X (predictor)and Y (response) are two matrix , I am regressing X ( dimensions 10 x 20) on each of columns of Y[,1] (10 x 1) and want to store the coefficient values. I have performed a Elastic Net regression and I want to store the coeffcients in each iteration. I got an error message . I do not know where is the problem Please help me. Thanks *Code:* --- library(elasticnet) X-matrix(rnorm(200),ncol=20) Y-matrix(rnorm(200),ncol=20) loop - 20 size - 20 enres-matrix(nrow = size, ncol = loop) fit-matrix(nrow = size, ncol = loop) store-matrix(nrow = size, ncol = loop) for(j in 1: 10) print (paste(j,/200,sep=)) { enres-enet(x=X,y=Y[,j],lambda=1,normalize=TRUE,intercept=TRUE) fit-predict.enet(enres, X, type=coefficients) store[,j]-fit$coefficients } library(elasticnet) Loading required package: lars X-matrix(rnorm(200),ncol=20) Y-matrix(rnorm(200),ncol=20) loop - 20 size - 20 enres-matrix(nrow = size, ncol = loop) fit-matrix(nrow = size, ncol = loop) store-matrix(nrow = size, ncol = loop) for(j in 1: 10) + print (paste(j,/200,sep=)) [1] 1/200 [1] 2/200 [1] 3/200 [1] 4/200 [1] 5/200 [1] 6/200 [1] 7/200 [1] 8/200 [1] 9/200 [1] 10/200 { + enres-enet(x=X,y=Y[,j],lambda=1,normalize=TRUE,intercept=TRUE) + fit-predict.enet(enres, X, type=coefficients) + store[,j]-fit$coefficients + } *Error in store[, j] - fit$coefficients : number of items to replace is not a multiple of replacement length * [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] Sparse PCA problem
Dear R user, I want to do sparse principal component analysis (spca). I am using elastic net package for this and spca() and the code is following from the example. My question is How can I decide the *K =? *and *para=c(7,4,4,1,1,1)) . So, here k=6 i.e the no of Principal Components. and each pcs say , * ** pc1 number of non zero loading is 7 pc2 number of non zero loading is 4 pc3 number of non zero loading is 4 pc4 number of non zero loading is 1 pc5 number of non zero loading is 1 pc6 number of non zero loading is 1 *How can I know in which pc,s how many non zero loadings will be? Any code??? One answer can be cross validation but I did not find in the package. * ** *Thanks for your help* *Code:* library(elasticnet) out2-spca(pitprops,*K=6*,type=Gram,sparse=varnum,trace=TRUE,* para=c(7,4,4,1,1,1)) *iterations 10 iterations 20 iterations 30 iterations 40 out2 Call: spca(x = pitprops, K = 6, para = c(7, 4, 4, 1, 1, 1), type = Gram, sparse = varnum, trace = TRUE) 6 sparse PCs Pct. of exp. var. : 28.2 13.9 13.1 7.4 6.8 6.3 Num. of non-zero loadings : 7 4 4 1 1 1 Sparse loadings PC1PC2PC3 PC4 PC5 PC6 topdiam -0.477 0.003 0.000 0 0 0 length -0.469 0.000 0.000 0 0 0 moist0.000 0.785 0.000 0 0 0 testsg 0.000 0.619 0.000 0 0 0 ovensg 0.180 0.000 0.656 0 0 0 ringtop 0.000 0.000 0.589 0 0 0 ringbut -0.290 0.000 0.470 0 0 0 bowmax -0.343 -0.029 -0.048 0 0 0 bowdist -0.414 0.000 0.000 0 0 0 whorls -0.383 0.000 0.000 0 0 0 clear0.000 0.000 0.000 -1 0 0 knots0.000 0.000 0.000 0 -1 0 diaknot 0.000 0.000 0.000 0 0 1 [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] Double Cross validation for LASSO
Dear R user, I am looking for a code on double cross validation in LASSO , one for optimizing the parameter and other one is for MSEP. If any one have it, please foroward to me. I am using different package like LARS, chemometric etc. Thanks in advance Alex [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Re: [R] Singularity in a regression?
If collinearity exists, one of the solutions is regulazation version of regression. There are different types of regularization method. like Ridge, LASSO, elastic net etc. For example, in MASS package you can get ridge regression. Alex On Thu, Feb 26, 2009 at 1:58 PM, Bob Gotwals gotw...@ncssm.edu wrote: R friends, In a matrix of 1s and 0s, I'm getting a singularity error. Any helpful ideas? lm(formula = activity ~ metaF + metaCl + metaBr + metaI + metaMe + paraF + paraCl + paraBr + paraI + paraMe) Residuals: Min 1Q Median 3QMax -4.573e-01 -7.884e-02 3.469e-17 6.616e-02 2.427e-01 Coefficients: (1 not defined because of singularities) Estimate Std. Error t value Pr(|t|) (Intercept) 7.9173 0.1129 70.135 2e-16 *** metaF-0.3973 0.2339 -1.698 0.115172 metaClNA NA NA NA metaBr0.3454 0.1149 3.007 0.010929 * metaI 0.4827 0.2339 2.063 0.061404 . metaMe0.3654 0.1149 3.181 0.007909 ** paraF 0.7675 0.1449 5.298 0.000189 *** paraCl0.3400 0.1449 2.347 0.036925 * paraBr1.0200 0.1449 7.040 1.36e-05 *** paraI 1.3327 0.2339 5.697 9.96e-05 *** paraMe1.2191 0.1573 7.751 5.19e-06 *** --- Signif. codes: 0 *** 0.001 ** 0.01 * 0.05 . 0.1 1 Residual standard error: 0.2049 on 12 degrees of freedom Multiple R-squared: 0.9257, Adjusted R-squared: 0.8699 F-statistic: 16.61 on 9 and 12 DF, p-value: 1.811e-05 __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.htmlhttp://www.r-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code. [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] Leave one out Cross validation (LOO)
Dear R user, I am working with LOO. Can any one who is working with leave one out cross validation (LOO) could send me the code? Thanks in advance Alex [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] e1071 package for SVM
Dear all, I got a code for e1071 package in R for SVM regression. I have used *m$coefs* for extracting the coefficients but I am getting only 72 . How can I extract coefficients of the predictors set? Does it mean that I will get only 72 as *Number of Support Vectors: 72. * ** Thanks in advance Code: -- library(e1071) # create data x - seq(0.1, 5, by = 0.05) y - log(x) + rnorm(x, sd = 0.2) # estimate model and predict input values m - svm(x, y) new - predict(m, x) m Call: svm.default(x = x, y = y) Parameters: SVM-Type: eps-regression SVM-Kernel: radial cost: 1 gamma: 1 epsilon: 0.1 *Number of Support Vectors: 72* * m$coefs* new 12345 6789 10 11 12 -1.327786564 -1.277059853 -1.221424097 -1.161313628 -1.097200621 -1.029588549 -0.959005127 -0.885994883 -0.81473 -0.734909901 -0.657938792 -0.580732849 13 14 15 16 17 18 19 20 21 22 23 24 -0.503805655 -0.427642943 -0.352696474 -0.279378612 -0.208057720 -0.139054438 -0.072638906 -0.009028968 0.051610615 0.109167970 0.163582945 0.214845022 25 26 27 28 29 30 31 32 33 34 35 36 0.262990384 0.308098215 0.350286330 0.389706259 0.426537887 0.460983809 0.493263495 0.523607428 0.552251329 0.579430594 0.605375062 0.630304214 37 38 39 40 41 42 43 44 45 46 47 48 0.654422894 0.677917633 0.700953619 0.723672382 0.746190188 0.768597174 0.790957190 0.813308346 0.835664193 0.858015504 0.880332573 0.902567958 49 50 51 52 53 54 55 56 57 58 59 60 0.924659570 0.946534037 0.968110220 0.989302798 1.010025830 1.030196181 1.049736754 1.068579418 1.086667584 1.103958361 1.120424239 1.136054277 61 62 63 64 65 66 67 68 69 70 71 72 1.150854762 1.164849331 1.178078576 1.190599126 1.202482259 1.213812069 1.224683251 1.235198555 1.245465988 1.255595825 1.265697515 1.275876563 73 74 75 76 77 78 79 80 81 82 83 84 1.286231447 1.296850678 1.307810045 1.319170135 1.330974176 1.343246256 1.355989966 1.369187496 1.382799201 1.396763660 1.410998202 1.425399923 85 86 87 88 89 90 91 92 93 94 95 96 1.439847127 1.454201198 1.468308834 1.482004603 1.495113765 1.507455297 1.518845050 1.529098984 1.538036397 1.545483098 1.551274451 1.555258224 97 98 99 1.557297215 1.557271572 1.555080800 [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] SVM regression code
Dear R user, I am looking for SVM regression in R. It willl be helpful for me if some one send me SVM regression code. Thanks Alex [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Re: [R] Subset Regression Package
Thank you very much for your help Alex On Wed, Feb 18, 2009 at 1:26 PM, Pedro Silva psi...@porto.ucp.pt wrote: -- Message: 72 Date: Tue, 17 Feb 2009 22:05:46 í (UTC) From: Hans W. Borchers hwborch...@gmail.com Subject: Re: [R] Subset Regression Package To: r-h...@stat.math.ethz.ch Message-ID: loom.20090217t215556-...@post.gmane.org Content-Type: text/plain; charset=us-ascii Take also a look at the subselect package that can perform subset selection in regression (and in several other statistical problems) using both exact (leaps and bounds algorithm) and heuristic (simulated annealing, genetic search, etc.) methods. Regards, A. Pedro Duarte Silva Alex Roy alexroy2008 at gmail.com writes: Dear all , Is there any subset regression (subset selection regression) package in R other than leaps? Lars and Lasso are other 'subset selection' methods, see the corresponding packages 'lars' and 'lasso2' and its description in The Elements of Statistical Learning. Also, 'dr', Methods for dimension reduction for regression, or 'relaimpo', Relative importance of regressors in linear models, can be considered. Thanks and regards Alex *** Esta mensagem (incluindo quaisquer anexos) pode conter informa豫o confidencial ou legalmente protegida para uso exclusivo do destinat�io. Se n� for o destinat�io pretendido da mesma, n� dever�fazer uso, copiar, distribuir ou revelar o seu conte�o (incluindo quaisquer anexos) a terceiros, sem a devida autoriza豫o. Se recebeu esta mensagem por engano, por favor informe o emissor, por e-mail, e elimine-a imediatamente. Obrigado. This message may contain confidential information or pri...{{dropped:6}} __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.htmlhttp://www.r-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code. [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] Subset Regression Package
Dear all , Is there any subset regression (subset selection regression) package in R other than leaps? Thanks and regards Alex [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] PRESS / RMSEP
Dear all , I want to do PRESS (prediction error sums of squares) or the residual mean square error of prediction (RMSEP) which will give me value that is valid for 'future predictions of independent data'. I am using different methods for example, Multiple Linear Regression, LASSO regression, Ridge Regression, Elastic Net regression etc. I am wandering if there are some package(s) in R or some websites/materials for such type of method. Thanks and regards Alex [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.