Hi,
I have huge matrices in which the response variable is in the first
column and the regressors are in the other columns. What I wanted to do
now is something like this:
#this is just to get an example-matrix
DataMatrix <- rep(1,1000);
Disturbance <- rnorm(900);
DataMatrix[101:1000] <- DataMatrix[101:1000]+Disturbance;
DataMatrix <- matrix(DataMatrix,ncol=10,nrow=100);
#estimate univariate linear model with each regressor-column, response
in the first column
for(i in 2:10){
result <- lm(DataMatrix[,1]~DataMatrix[,i])
}
Is there any way to get rid of the for-loop using mapply (or some other
function)?
Thanks!
Philipp
On Sep 8, 2010, at 7:34 AM, Philipp Kunze wrote:> Hi, > I have huge matrices in which the response variable is in the first > column and the regressors are in the other columns. What I wanted to > do > now is something like this: > > #this is just to get an example-matrix > DataMatrix <- rep(1,1000); > Disturbance <- rnorm(900); > DataMatrix[101:1000] <- DataMatrix[101:1000]+Disturbance; > DataMatrix <- matrix(DataMatrix,ncol=10,nrow=100); > > #estimate univariate linear model with each regressor-column, response > in the first column > > for(i in 2:10){ > result <- lm(DataMatrix[,1]~DataMatrix[,i]) > }result <- apply(DataMatrix[,2:10], 2, function (x) lm(DataMatrix[, 1]~x) ) Which would have the added advantage that "result" would not be overwritten for iterations 3:10, which is what your code would have done. "result" will be a list of 9 models which might be a bit unweildy, so you might consider something like result <- apply(DataMatrix[,2:10], 2, function (x) coef( lm(DataMatrix[,1]~x) ) ) result When you do so, you uncover a fatal flaw in your strategy, which suggests you have not even done this once on your data or simulations. -- David.> > > Is there any way to get rid of the for-loop using mapply (or some > other > function)?-- David Winsemius, MD West Hartford, CT
Hi Philipp, I like to use something like lapply(2:10, function(j) lm.fit(cbind(1, DataMatrix[,j]), DataMatrix[,1])) for this sort of thing. I'd be curious to know if there are other approaches that are better. --Gray On Wed, Sep 8, 2010 at 4:34 AM, Philipp Kunze <pkunze at gwdg.de> wrote:> Hi, > I have huge matrices in which the response variable is in the first > column and the regressors are in the other columns. What I wanted to do > now is something like this: > > #this is just to get an example-matrix > DataMatrix <- rep(1,1000); > Disturbance <- rnorm(900); > DataMatrix[101:1000] <- DataMatrix[101:1000]+Disturbance; > DataMatrix <- matrix(DataMatrix,ncol=10,nrow=100); > > #estimate univariate linear model with each regressor-column, response > in the first column > > for(i in 2:10){ > ? ? ? ?result <- lm(DataMatrix[,1]~DataMatrix[,i]) > } > > > Is there any way to get rid of the for-loop using mapply (or some other > function)? > > Thanks! > Philipp > > ______________________________________________ > R-help at 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. >-- Gray Calhoun Assistant Professor of Economics, Iowa State University http://www.econ.iastate.edu/~gcalhoun/
This really depends on why you want to do this and what results you want. If your main goal is to look at some basic tests, goodness of fit, then the add1 function may do everything you need. If you just want coefficient estimates then some basic matrix algebra will give those to you. Another option would be to reshape the data to long format and use lmList from the nlme package (the above will be quicker if you do not need everything that lm gives you). -- Gregory (Greg) L. Snow Ph.D. Statistical Data Center Intermountain Healthcare greg.snow at imail.org 801.408.8111> -----Original Message----- > From: r-help-bounces at r-project.org [mailto:r-help-bounces at r- > project.org] On Behalf Of Philipp Kunze > Sent: Wednesday, September 08, 2010 5:35 AM > To: R-help at r-project.org > Subject: [R] Regression using mapply? > > Hi, > I have huge matrices in which the response variable is in the first > column and the regressors are in the other columns. What I wanted to do > now is something like this: > > #this is just to get an example-matrix > DataMatrix <- rep(1,1000); > Disturbance <- rnorm(900); > DataMatrix[101:1000] <- DataMatrix[101:1000]+Disturbance; > DataMatrix <- matrix(DataMatrix,ncol=10,nrow=100); > > #estimate univariate linear model with each regressor-column, response > in the first column > > for(i in 2:10){ > result <- lm(DataMatrix[,1]~DataMatrix[,i]) > } > > > Is there any way to get rid of the for-loop using mapply (or some other > function)? > > Thanks! > Philipp > > ______________________________________________ > R-help at 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.