Wentzel-Larsen, Tore
2007-Aug-27 13:58 UTC
[R] validate (package Design): error message "subscript out of bounds"
Dear R users
I use Windows XP, R2.5.1 (I have read the posting guide, I have
contacted the package maintainer first, it is not homework).
In a research project on renal cell carcinoma we want to compute
Harrell's c index, with optimism correction, for a multivariate
Cox regression and also for some univariate Cox models.
For some of these univariate models I have encountered an error
message (and no result produced) from the function validate i
Frank Harrell's Design package:
Error in Xb(x[, xcol, drop = FALSE], coef, non.slopes, non.slopes.in.x, :
subscript out of bounds
The following is an artificial example wherein I have been able to
reproduce this error message (actual data has been changed to preserve
confidentiality):
library(Design)
# an example data frame:
frame.bc <- data.frame(time1 = c(9,24,28,43,58,62,66,107,116,118,123,
127,129,131,137,138,139,140,148,169,176,179,188,196,210,218,
1,1,1,2,2,3,4,8,23,32,33,34,43,44,48,51,52,54,59,59,60,60,62,
65,65,68,70,72,73,74,81,84,88,98,99,106,107,115,115,117,119,
120,122,122,122,122,126,128,130,135,136,136,138,149,151,154,
157,159,161,164,164,164,166,172,172,176,179,180,183,183,184,
187,190,197,201,201,203,203,203,209,210,214,219,227,233,4,18,
49,113,147,1,1,2,2,2,2,2,3,4,6,6,6,6,6,6,6,6,9,9,9,9,9,10,10,
10,11,12,12,12,13,14,14,17,18,18,19,19,20,20,21,21,21,21,22,23,
23,24,28,28,29,29,32,34,35,38,38,48,48,52,52,54,54,56,64,67,67,
69,70,70,72,84,88,90,114,115,140,142,154,171,195),
status1 = c(0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
1,1,1,1,1),
bc1 = factor(c(1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,
2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,
2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,1,1,1,1,1,2,2,2,2,2,2,2,2,2,2,2,2,
2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,
2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2),
labels=c('bc.1','bc.2')),
age = c(58,68,23,20,50,43,41,69,20,48,19,27,39,20,65,49,70,59,31,43,25,
61,60,45,34,59,32,58,30,62,26,44,52,29,40,57,33,18,50,50,55,51,38,34,
69,56,67,38,66,21,48,39,62,62,29,68,66,19,60,39,55,42,24,29,56,61,40,
52,19,40,33,67,66,51,48,63,60,58,68,60,53,20,45,62,37,38,61,63,43,67,
49,39,43,67,49,69,32,37,32,63,33,47,66,39,23,57,26,61,20,49,69,30,40,
29,38,66,60,69,69,44,65,25,41,53,18,55,45,59,49,27,51,29,67,26,24,26,
47,23,50,27,35,45,32,26,45,45,63,39,39,22,38,27,31,27,49,65,66,49,39,
21,51,49,55,63,19,26,50,21,24,34,65,33,55,33,36,53,48,25,54,58,60,34,
47,23,34,60,39,34,22,30,41,55,64,48,34,54))
frame.bc
# preparing for a simple univariate Cox regression:
dd.bc <- datadist(frame.bc[, c('bc1','age')],
adjto.cat='first')
options(datadist = 'dd.bc')
# a univariate Cox regression:
cph.bc <- cph(formula = Surv(time1,status1)~bc1,
data = frame.bc, x=TRUE, y=TRUE, surv=TRUE)
anova(cph.bc)
cph.bc
summary(cph.bc)
# the validate command for the Cox model:
val.cph.bc <- validate(cph.bc, B=200, dxy=TRUE , pr=TRUE)
----------------------
Output from the validate command:
training test
Dxy -0.124360 -0.1423409
R2 1.000000 1.0000000
Slope 1.000000 0.7919584
D 0.016791 0.0147536
U -0.002395 0.0006448
Q 0.019186 0.0141088
training test
Dxy -0.191875 -0.1423409
R2 1.000000 1.0000000
Slope 1.000000 0.8936724
D 0.022397 0.0147536
U -0.002339 0.0001367
Q 0.024736 0.0146169
training test
Dxy -0.199514 -0.1423409
R2 1.000000 1.0000000
Slope 1.000000 0.8075246
D 0.025717 0.0147536
U -0.002447 0.0005348
Q 0.028163 0.0142188
Error in Xb(x[, xcol, drop = FALSE], coef, non.slopes, non.slopes.in.x, :
subscript out of bounds
Any help/suggestions will be highly appreciated.
Sincerely,
Tore Wentzel-Larsen
statistician
Centre for Clinical research
Armauer Hansen house
Haukeland University Hospital
N-5021 Bergen
tlf +47 55 97 55 39 (a)
faks +47 55 97 60 88 (a)
email tore.wentzel-larsen at helse-bergen.no
Frank E Harrell Jr
2007-Aug-28 01:16 UTC
[R] validate (package Design): error message "subscript out of bounds"
Wentzel-Larsen, Tore wrote:> Dear R users > > I use Windows XP, R2.5.1 (I have read the posting guide, I have > contacted the package maintainer first, it is not homework). > > In a research project on renal cell carcinoma we want to compute > Harrell's c index, with optimism correction, for a multivariate > Cox regression and also for some univariate Cox models. > For some of these univariate models I have encountered an error > message (and no result produced) from the function validate i > Frank Harrell's Design package: > > Error in Xb(x[, xcol, drop = FALSE], coef, non.slopes, non.slopes.in.x, : > subscript out of bounds > > The following is an artificial example wherein I have been able to > reproduce this error message (actual data has been changed to preserve > confidentiality):I could not reproduce the error on R 2.5.1 on linux using version 2.0-12 of Design (you did not provide this information). Your code involved a good deal of extra typing. Here is a streamlined version: bc <- data.frame(time1 = c(9,24,28,43,58,62,66,107,116,118,123, 127,129,131,137,138,139,140,148,169,176,179,188,196,210,218, .... bc library(Design) dd <- with(bc, datadist(bc1, age, adjto.cat='first')) options(datadist = 'dd') f <- cph(Surv(time1,status1) ~ bc1, data = bc, x=TRUE, y=TRUE, surv=TRUE) anova(f) f summary(f) val <- validate(f, B=200, dxy=TRUE) I don't get much value of putting the type of an object as part of the object's name, as information within objects defines the object type/class. There is little reason to validate a one degree of freedom model. Frank> > library(Design) > > # an example data frame: > frame.bc <- data.frame(time1 = c(9,24,28,43,58,62,66,107,116,118,123, > 127,129,131,137,138,139,140,148,169,176,179,188,196,210,218, > 1,1,1,2,2,3,4,8,23,32,33,34,43,44,48,51,52,54,59,59,60,60,62, > 65,65,68,70,72,73,74,81,84,88,98,99,106,107,115,115,117,119, > 120,122,122,122,122,126,128,130,135,136,136,138,149,151,154, > 157,159,161,164,164,164,166,172,172,176,179,180,183,183,184, > 187,190,197,201,201,203,203,203,209,210,214,219,227,233,4,18, > 49,113,147,1,1,2,2,2,2,2,3,4,6,6,6,6,6,6,6,6,9,9,9,9,9,10,10, > 10,11,12,12,12,13,14,14,17,18,18,19,19,20,20,21,21,21,21,22,23, > 23,24,28,28,29,29,32,34,35,38,38,48,48,52,52,54,54,56,64,67,67, > 69,70,70,72,84,88,90,114,115,140,142,154,171,195), > status1 = c(0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, > 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, > 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, > 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,1,1,1,1,1,1,1, > 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, > 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, > 1,1,1,1,1), > bc1 = factor(c(1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, > 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, > 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, > 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,1,1,1,1,1,2,2,2,2,2,2,2,2,2,2,2,2, > 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, > 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2), > labels=c('bc.1','bc.2')), > age = c(58,68,23,20,50,43,41,69,20,48,19,27,39,20,65,49,70,59,31,43,25, > 61,60,45,34,59,32,58,30,62,26,44,52,29,40,57,33,18,50,50,55,51,38,34, > 69,56,67,38,66,21,48,39,62,62,29,68,66,19,60,39,55,42,24,29,56,61,40, > 52,19,40,33,67,66,51,48,63,60,58,68,60,53,20,45,62,37,38,61,63,43,67, > 49,39,43,67,49,69,32,37,32,63,33,47,66,39,23,57,26,61,20,49,69,30,40, > 29,38,66,60,69,69,44,65,25,41,53,18,55,45,59,49,27,51,29,67,26,24,26, > 47,23,50,27,35,45,32,26,45,45,63,39,39,22,38,27,31,27,49,65,66,49,39, > 21,51,49,55,63,19,26,50,21,24,34,65,33,55,33,36,53,48,25,54,58,60,34, > 47,23,34,60,39,34,22,30,41,55,64,48,34,54)) > frame.bc > > # preparing for a simple univariate Cox regression: > dd.bc <- datadist(frame.bc[, c('bc1','age')], adjto.cat='first') > options(datadist = 'dd.bc') > > # a univariate Cox regression: > cph.bc <- cph(formula = Surv(time1,status1)~bc1, > data = frame.bc, x=TRUE, y=TRUE, surv=TRUE) > anova(cph.bc) > cph.bc > summary(cph.bc) > > # the validate command for the Cox model: > val.cph.bc <- validate(cph.bc, B=200, dxy=TRUE , pr=TRUE) > > ---------------------- > Output from the validate command: > > training test > Dxy -0.124360 -0.1423409 > R2 1.000000 1.0000000 > Slope 1.000000 0.7919584 > D 0.016791 0.0147536 > U -0.002395 0.0006448 > Q 0.019186 0.0141088 > training test > Dxy -0.191875 -0.1423409 > R2 1.000000 1.0000000 > Slope 1.000000 0.8936724 > D 0.022397 0.0147536 > U -0.002339 0.0001367 > Q 0.024736 0.0146169 > training test > Dxy -0.199514 -0.1423409 > R2 1.000000 1.0000000 > Slope 1.000000 0.8075246 > D 0.025717 0.0147536 > U -0.002447 0.0005348 > Q 0.028163 0.0142188 > Error in Xb(x[, xcol, drop = FALSE], coef, non.slopes, non.slopes.in.x, : > subscript out of bounds > > > Any help/suggestions will be highly appreciated. > > > Sincerely, > Tore Wentzel-Larsen > statistician > Centre for Clinical research > Armauer Hansen house > Haukeland University Hospital > N-5021 Bergen > tlf +47 55 97 55 39 (a) > faks +47 55 97 60 88 (a) > email tore.wentzel-larsen at helse-bergen.no > > ______________________________________________ > R-help at stat.math.ethz.ch 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. >-- Frank E Harrell Jr Professor and Chair School of Medicine Department of Biostatistics Vanderbilt University
Wentzel-Larsen, Tore
2007-Aug-28 07:31 UTC
[R] validate (package Design): error message "subscript out of bounds"
Thanks, I use Design version 2.1-1 (and as provided, R 2.5.1 on Windows XP). The redundancies in object names were for my own convience, as part of a larger command file, and the validation of this univariate model was only included to ease comparison with the main multivariate model. I have tried to access Design version 2.0_12, but only managed to access version 2.1_1 of Design in my Windows implementation of R2.5.1 (the choice of operative system is made by my institution and I am only entitled to use Windows). Best, Tore -----Opprinnelig melding----- Fra: Frank E Harrell Jr [mailto:f.harrell at vanderbilt.edu] Sendt: 28. august 2007 03:17 Til: Wentzel-Larsen, Tore Kopi: r-help at stat.math.ethz.ch Emne: Re: [R] validate (package Design): error message "subscript out of bounds" Wentzel-Larsen, Tore wrote:> Dear R users > > I use Windows XP, R2.5.1 (I have read the posting guide, I have > contacted the package maintainer first, it is not homework). > > In a research project on renal cell carcinoma we want to compute > Harrell's c index, with optimism correction, for a multivariate > Cox regression and also for some univariate Cox models. > For some of these univariate models I have encountered an error > message (and no result produced) from the function validate i > Frank Harrell's Design package: > > Error in Xb(x[, xcol, drop = FALSE], coef, non.slopes, non.slopes.in.x, : > subscript out of bounds > > The following is an artificial example wherein I have been able to > reproduce this error message (actual data has been changed to preserve > confidentiality):I could not reproduce the error on R 2.5.1 on linux using version 2.0-12 of Design (you did not provide this information). Your code involved a good deal of extra typing. Here is a streamlined version: bc <- data.frame(time1 = c(9,24,28,43,58,62,66,107,116,118,123, 127,129,131,137,138,139,140,148,169,176,179,188,196,210,218, .... bc library(Design) dd <- with(bc, datadist(bc1, age, adjto.cat='first')) options(datadist = 'dd') f <- cph(Surv(time1,status1) ~ bc1, data = bc, x=TRUE, y=TRUE, surv=TRUE) anova(f) f summary(f) val <- validate(f, B=200, dxy=TRUE) I don't get much value of putting the type of an object as part of the object's name, as information within objects defines the object type/class. There is little reason to validate a one degree of freedom model. Frank> > library(Design) > > # an example data frame: > frame.bc <- data.frame(time1 = c(9,24,28,43,58,62,66,107,116,118,123, > 127,129,131,137,138,139,140,148,169,176,179,188,196,210,218, > 1,1,1,2,2,3,4,8,23,32,33,34,43,44,48,51,52,54,59,59,60,60,62, > 65,65,68,70,72,73,74,81,84,88,98,99,106,107,115,115,117,119, > 120,122,122,122,122,126,128,130,135,136,136,138,149,151,154, > 157,159,161,164,164,164,166,172,172,176,179,180,183,183,184, > 187,190,197,201,201,203,203,203,209,210,214,219,227,233,4,18, > 49,113,147,1,1,2,2,2,2,2,3,4,6,6,6,6,6,6,6,6,9,9,9,9,9,10,10, > 10,11,12,12,12,13,14,14,17,18,18,19,19,20,20,21,21,21,21,22,23, > 23,24,28,28,29,29,32,34,35,38,38,48,48,52,52,54,54,56,64,67,67, > 69,70,70,72,84,88,90,114,115,140,142,154,171,195), > status1 = c(0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, > 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, > 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, > 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,1,1,1,1,1,1,1, > 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, > 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, > 1,1,1,1,1), > bc1 = factor(c(1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, > 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, > 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, > 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,1,1,1,1,1,2,2,2,2,2,2,2,2,2,2,2,2, > 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, > 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2), > labels=c('bc.1','bc.2')), > age = c(58,68,23,20,50,43,41,69,20,48,19,27,39,20,65,49,70,59,31,43,25, > 61,60,45,34,59,32,58,30,62,26,44,52,29,40,57,33,18,50,50,55,51,38,34, > 69,56,67,38,66,21,48,39,62,62,29,68,66,19,60,39,55,42,24,29,56,61,40, > 52,19,40,33,67,66,51,48,63,60,58,68,60,53,20,45,62,37,38,61,63,43,67, > 49,39,43,67,49,69,32,37,32,63,33,47,66,39,23,57,26,61,20,49,69,30,40, > 29,38,66,60,69,69,44,65,25,41,53,18,55,45,59,49,27,51,29,67,26,24,26, > 47,23,50,27,35,45,32,26,45,45,63,39,39,22,38,27,31,27,49,65,66,49,39, > 21,51,49,55,63,19,26,50,21,24,34,65,33,55,33,36,53,48,25,54,58,60,34, > 47,23,34,60,39,34,22,30,41,55,64,48,34,54)) > frame.bc > > # preparing for a simple univariate Cox regression: > dd.bc <- datadist(frame.bc[, c('bc1','age')], adjto.cat='first') > options(datadist = 'dd.bc') > > # a univariate Cox regression: > cph.bc <- cph(formula = Surv(time1,status1)~bc1, > data = frame.bc, x=TRUE, y=TRUE, surv=TRUE) > anova(cph.bc) > cph.bc > summary(cph.bc) > > # the validate command for the Cox model: > val.cph.bc <- validate(cph.bc, B=200, dxy=TRUE , pr=TRUE) > > ---------------------- > Output from the validate command: > > training test > Dxy -0.124360 -0.1423409 > R2 1.000000 1.0000000 > Slope 1.000000 0.7919584 > D 0.016791 0.0147536 > U -0.002395 0.0006448 > Q 0.019186 0.0141088 > training test > Dxy -0.191875 -0.1423409 > R2 1.000000 1.0000000 > Slope 1.000000 0.8936724 > D 0.022397 0.0147536 > U -0.002339 0.0001367 > Q 0.024736 0.0146169 > training test > Dxy -0.199514 -0.1423409 > R2 1.000000 1.0000000 > Slope 1.000000 0.8075246 > D 0.025717 0.0147536 > U -0.002447 0.0005348 > Q 0.028163 0.0142188 > Error in Xb(x[, xcol, drop = FALSE], coef, non.slopes, non.slopes.in.x, : > subscript out of bounds > > > Any help/suggestions will be highly appreciated. > > > Sincerely, > Tore Wentzel-Larsen > statistician > Centre for Clinical research > Armauer Hansen house > Haukeland University Hospital > N-5021 Bergen > tlf +47 55 97 55 39 (a) > faks +47 55 97 60 88 (a) > email tore.wentzel-larsen at helse-bergen.no > > ______________________________________________ > R-help at stat.math.ethz.ch 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. >-- Frank E Harrell Jr Professor and Chair School of Medicine Department of Biostatistics Vanderbilt University