similar to: [OT] - standard errors for parameter estimates under ridge regression and lasso?

Dear community, I'm trying to track down a quote, but can't recall the source or the exact structure - not very helpful, I know - something along the lines that: 80% of [applied] statistics is linear regression ... ? Does this ring a bell for anyone? Thanks, Andrew -- Andrew Robinson Department of Mathematics and Statistics Tel: +61-3-8344-9763 University of

Dear R-helpers, when in the browser, is it possible to change the environment, so as to be able to easily access (print, manipulate) objects in the parent, or elsehwere? I know that it is possible to evaluate expressions in different environments, using eval(), but I would prefer to avoid that if possible. Thanks, Andrew -- Andrew Robinson Department of Mathematics and Statistics

Dear all, I'm curious about how to interpret the results of the following code. The first model is directly from the help page of lmer; the second is the same model but using the Gamma family with log link. The fixed effects make sense, because y = 251.40510 + 10.46729 * Days is about the same as log(y) = 5.53613298 + 0.03502057 * Days but the random effects seem quite

Dear Friends and Colleagues, I'm puzzling over how to interpret or use some bootstrap intervals. I think that I know what I should do, but I want to check with knowledgeable people first! I'm using a studentized non-parametric bootstrap to estimate 95% confidence intervals for three parameters. I estimate the variance of the bootstrap replicates using another bootstrap. The script

Hi Community, I'm trying to reconcile Cook's Distances computed in glm. The following snippet of code shows that the Cook's Distances contours on the plot of Residuals v Leverage do not seem to be the same as the values produced by cooks.distance() or in the Cook's Distance against observation number plot. counts <- c(18,17,15,20,10,20,25,13,12) outcome <- gl(3,1,9)

Dear R-developers, when tapply() is invoked on factors that have empty levels, it returns NA. This behaviour is in accord with the tapply documentation, and is reasonable in many cases. However, when FUN is sum, it would also seem reasonable to return 0 instead of NA, because "the sum of an empty set is zero, by definition." I'd like to raise a discussion of the possibility of an

Dear R colleagues, a friend and I are trying to develop a modest workflow for the problem of decomposing tests of higher-order terms into interpretable sets of tests of lower order terms with conditioning. For example, if the interaction between A (3 levels) and C (2 levels) is significant, it may be of interest to ask whether or not A is significant at level 1 of C and level 2 of C. The

Dear R-users, we are happy to announce the release of our R package parcor. The package contains tools to estimate the matrix of partial correlations based on different regularized regression methods: Lasso, adaptive Lasso, PLS, and Ridge Regression. In addition, parcor provides cross-validation based model selection for Lasso, adaptive Lasso and Ridge Regression. More details can be found

Dear R-users, we are happy to announce the release of our R package parcor. The package contains tools to estimate the matrix of partial correlations based on different regularized regression methods: Lasso, adaptive Lasso, PLS, and Ridge Regression. In addition, parcor provides cross-validation based model selection for Lasso, adaptive Lasso and Ridge Regression. More details can be found

Dear All The problem is about regularization methods in multiple regression when the independent variables are collinear. A modified regularization method with two tuning parameters l1 and l2 and their product l1*l2 (Lambda 1 and Lambda 2) such that l1 takes care of ridge property and l2 takes care of LASSO property is proposed The proposed method is given

Hi everyone, I'm involved in a discussion with a colleague. He suggested a sample design for a finite-sized process that (to all intents and purposes) involves tossing a coin and examining the unit if the coin shows Heads. I should emphasize that we're both approaching the problem from a design-based sampling theory point of view. So I have no argument about the appropriateness of the

Hi, Can anyone explain how the standard error in arima() is calculated? Also, how can I extract it from the Arima object? I don't see it in there. > x <- rnorm(1000) > a <- arima(x, order = c(4, 0, 0)) > a Call: arima(x = x, order = c(4, 0, 0)) Coefficients: ar1 ar2 ar3 ar4 intercept -0.0451 0.0448 0.0139 -0.0688 0.0010 s.e.

Dear R community, I'm trying to figure out if there is any way to split the printed output of some commands, for example summary.lme, so that I can intersperse comments in Sweave. I don't mind running the command numerous times and masking various portions of the output, or saving the output as an object and printing it, but I can't figure out how to do either. Does anyone have any

Hi everyone, I was wondering if anyone has any code they could share for creating thumbnail plots in Sweave. For example, I'd like a plot like the following: y <- c(40, 46, 39, 44, 23, 36, 70, 39, 30, 73, 53, 74) x <- c(6, 4, 3, 6, 1, 5, 6, 2, 1, 8, 4, 6) opar <- par(mar=c(3,3,0,0)) plot(x, y, xlab="", ylab="") abline(h=mean(y), col="red") par(opar)

Dear community, can anyone provide a snippet of code to force the lm() to fit a model with terms in the formula in an arbitrary order? I am interested in something like: lm(y ~ A * B + C, data=data) where the interaction of A and B should be in the formula before C. My goal is to simplify my presentation of models using the anova() statement. I have found that this should be possible using

As a fourth option, I wonder if the first author could fork the package? Presumably, appropriately cited, a fork is permitted by the license under which it was released. Then the original package, by both authors, still exists (and a final version could point to the new one) and the new package, citing the previous version appropriately, is by a single author. The page of CRAN's policies

Unlike L1 (lasso) regression or elastic net (mixture of L1 and L2), L2 norm regression (ridge regression) does not select variables. Selection of variables would not work properly, and it's unclear why you would want to omit "apparently" weak variables anyway. Frank maths123 wrote > I have a .txt file containing a dataset with 500 samples. There are 10 > variables. > >

Hi R-helpers, I am trying to build a package under FreeBSD 5.4-RELEASE #0 using R Version 2.1.1. I have constructed a package using package.skeleton(), when I try $ R CMD build foo * checking for file 'foo/DESCRIPTION' ... OK * preparing 'foo': * checking DESCRIPTION meta-information ... OK * cleaning src * removing junk files tar: Option -L is not permitted in mode -x Error:

Dear Community, I'm interested in developing a package that could ease the command-line learning curve for new users. It would provide more detailed syntax checking and commentary as feedback. It would try to anticipate common new-user errors, and provide feedback to help correct them. As a trivial example, instead of > mean(c(1,2,NA)) [1] NA we might have > mean(c(1,2,NA)) [1]

The Rassist package has been loaded to CRAN. This package is designed to make R easier for new users, by providing extra checks and feedback. Presently the package functionality includes: * offers an alternative help facility, eg(.), with examples first, with additional examples included. eg() offers a start help menu, and eg(.) incorporates help.search(.) automatically. It also