Skip to content

save the output of summary(lmList(x)) into a dataframe

5 messages · Cecilia Carmo, milton ruser, Daniel Malter +1 more

Original
Cecilia Carmo
# 
Hi r-helpers!

I need to save the output of summary() function that I?ve 
runned like this:
z<- lmList(y~x1+x2| x3, 
na.action=na.omit,data1,subset=year==1999)
w<-summary(z)
The output (w) is something like this:
Call:
   Model: y ~ x1 + x2 | x3
    Data: data1
Coefficients:
    (Intercept)
         Estimate    Std. Error       t value  Pr(>|t|)
1    0.081110514  1.141352e-01  7.106531e-01 0.4774081
2    0.384008008           NaN           NaN       NaN
5   -0.127962259           NaN           NaN       NaN
14   0.828115409  6.888789e-01  5.427381e-02 0.9567241


x1
         Estimate    Std. Error       t value  Pr(>|t|)
1    58.06999452 137.263534154  4.230548e-01 0.6723151
2  -202.45105435           NaN           NaN       NaN
5            NaN           NaN           NaN       NaN
14    0.12776835   0.047486161  1.262510e-02 0.9899286


x2
        Estimate    Std. Error       t value     Pr(>|t|)
1  -0.173986693  2.126186e-01 -8.183043e-01 4.133115e-01
2  -0.630343567           NaN           NaN          NaN
5           NaN           NaN           NaN          NaN
14  0.737871169 -3.203230e+01  3.840265e+01 0.000000e+00


Residual standard error: 0.2261486 on 1522 degrees of 
freedom

I would like to have this output  in a dataframe like 
this:
x3  intercept  x1   x2   Std. Error   t value   Pr(>|t|)
1           

2
5
14



Anyone knows if it is possible and how?

Thanks in advance,

Cec?lia Carmo (Universidade de Aveiro ? Portugal)
Cecilia Carmo
# 
I'm sending the dataframe with my data and the function I 
have runned.

regaccdis<-read.table(file="regaccdis.txt", sep="\t", 
header=TRUE)
w<- lmList(totaccz~lactivoz+varvolnegz| caedois, 
na.action=na.omit,regaccdis,subset=ano==1999)
z<-summary(w)

Thank you all,
Cec?lia


Em Tue, 16 Jun 2009 09:56:26 -0400
  milton ruser <milton.ruser at gmail.com> escreveu:
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: regaccdis.txt
URL: <https://stat.ethz.ch/pipermail/r-help/attachments/20090616/03a93531/attachment-0002.txt>
Daniel Malter
# 
Hi, your problem has nothing to do with your specific dataset. Further, it
is - let's say - suboptimal to push a 5 MB dataset through the mailing list.
The posting guides asks to provide an example with self-contained code; this
is what Milton has asked for. Such an example is not out of reach for your
problem (see below)...

Let's say we have 3 subjects (i in {1,2,3}). And they have y(i)=10*i+x*i+e,
where x and e and standard-normal. Then you run lmList, the results of which
are stored in what is called "reg" in this example. You can assess every
element of summary(reg) individually by specifying summary(reg)$coef[n,j,k],
where n, j, and k are index numbers. See the example below. Then you have to
create a matrix that arranges the coefficients and standard errors and what
not in the way you want. For efficiency, you would do this in a loop or
something similar. 

However, given the description of the resulting matrix you desire, it is
unclear which SE, t-value, and p-value you would like at the end of each
line of your results matrix. Remember that each of the coefficients has its
own SE, t-value, and p-value. Thus, your desired results matrix seems partly
nonsensical. To just get the coefficients (not SEs, t-values, and p-values)
type "reg" in the prompt (or for your own example just type "z" and not
"summary(z)"). From there you can copy the coefficients directly into MS
Excel, for example.

i=rep(1:3,each=100)
n=100
x=rnorm(n*i,0,1)
e=rnorm(n*i,0,1)
y=10*i+x*i+e
mydata=data.frame(y,x,i)

library(nlme)
reg=lmList(y~x|i,data=mydata)
summary(reg)

summary(reg)$coef[1,1,1] #intercept 1
summary(reg)$coef[2,1,1] #intercept 2
summary(reg)$coef[3,1,1] #intercept 3
summary(reg)$coef[1,1,2] #slope 1
summary(reg)$coef[2,1,2] #slope 2
summary(reg)$coef[3,1,2] #slope 3

summary(reg)$coef[1,2,1] #SE intercept 1
summary(reg)$coef[2,2,1] #SE intercept 2
summary(reg)$coef[3,2,1] #SE intercept 3
summary(reg)$coef[1,2,2] #SE slope 1
summary(reg)$coef[1,2,2] #SE slope 2
summary(reg)$coef[1,2,2] #SE slope 3

#etc.

Best,
Daniel

-------------------------
cuncta stricte discussurus
-------------------------

-----Urspr?ngliche Nachricht-----
Von: r-help-bounces at r-project.org [mailto:r-help-bounces at r-project.org] Im
Auftrag von Cecilia Carmo
Gesendet: Tuesday, June 16, 2009 10:33 AM
An: milton ruser
Cc: r-help at r-project.org
Betreff: Re: [R] save the output of summary(lmList(x)) into a dataframe

I'm sending the dataframe with my data and the function I have runned.

regaccdis<-read.table(file="regaccdis.txt", sep="\t",
header=TRUE)
w<- lmList(totaccz~lactivoz+varvolnegz| caedois,
na.action=na.omit,regaccdis,subset=ano==1999)
z<-summary(w)

Thank you all,
Cec?lia


Em Tue, 16 Jun 2009 09:56:26 -0400
  milton ruser <milton.ruser at gmail.com> escreveu:
Residual
Paul Johnson
#
On Tue, Jun 16, 2009 at 3:29 AM, Cecilia Carmo<cecilia.carmo at ua.pt> wrote:
Does this come close?  I'm just using the lmList example from lme4

### example(lmList) spawns an object "fm1" that I demonstrate with:


library(lme4)

example(lmList)


varnames <- names(fm1[[1]]$coefficients)



mylist <- lmlist(fm1, summary)

varnames <- names(fm1[[1]]$coefficients)


myCoefficients <- function(x, name) { x[name,] }

theintercepts <- t( sapply(mylist, myCoefficients, varnames[1]))

theSecondVar <- t( sapply(mylist, myCoefficients, varnames[2]))


I know other r-help readers will know a better way to cycle through
the variables rather than running once for each element of varnames...