Several times I had to correct database tables with a character primary key (which is ok, if not all other tables had numeric …). In order to get the necessary numeric format I inserted another variable with the correct format and transformed the character variable to numeric using the CDbl() Function.
The syntax for the CDbl function is:
CDbl([MyString])
I did forget it again and could not find any documentation. Googling the topic gave a lot of fruitless hits, so here it is.
Usually a Cox-regression is achieved in R by
library(survival)
model <- coxph ( SuvivalObject ~ Covariate1 + Covariate2 + Factor1 + Factor2 , data = Dataset )
The covariates can be enclosed in other funtions:
In the latter case the model might look like
model <- coxph (SurvivalObject ~ pspline(Covariate1) + Covariate2 + factor(Factor1) + strata(Factor2) , data = Dataset )
The functional form of the covariates (including the factors) can now be plotted with
termplot(model)
Though the termplot() function fails with plotting just one covariate and leaves no cusomization.
The function plotHR() plots the functional form of the desired term: plotHR(model)
plots the first term in the model by default but other terms can be accessed by calling their number (e.g. the second one):
plotHR(model , terms = 2)
In order to use the function you have to “source” it into R. It is the same procedure as calling a package, but using “source” instead of “library”.
Paste the function syntax into a textfile and safe it (as plot.HR.R) on your harddisk, remember the path and include
source("C:Path/to/plotHR_0.6.R")
before using the function.
Note: I have rewritten the function several times since I wrote the initial post … using version numbers now …
lines( h = 0 , type = 2 )
I rewrote the “rugs” option. Try rugs = "density" It is still “beta”ish, but some like it.
Usage:
plotHR( model , terms = 1 , se = TRUE , rug = "ticks" , x.log = FALSE , xlab = "" , ylab = "Hazard Ratio" , main = NULL , xlim = NULL , ylim = NULL, col.term = "#08519C", lwd.term = 3, col.se = "#DEEBF7", cex = 1 , bty = "n" , axes = TRUE )
?plot.default
Just playing around the other day to get the default plot.density() function a bit more like publishing quality. Above is my favorite so far. Further down are two more spartanic versions.
In this connection I also wrote my first function. When I get more customed to R I will package my ideas into an R library – but later…
There is the possibility to call R-functions without embedding the code into your analysis script, but I did not look that up yet, so embedding the following code into your script (at the beginning) and then using
densityplot(Dataset$Coviate)
will do the job, where Dataset and Covariate have to be replaced by the according values of course.
Now the function code:
# function densityplot
densityplot <- function(x , digits = 1 , xlab = "" , ylab = "" , main = "Density" , col = "blue"){
dens <- density(x , na.rm = T)
bins <- as.numeric(cut(dens$x , breaks = fivenum(x)))
i1 <- bins == 1 & !is.na(bins)
i2 <- bins == 2 & !is.na(bins)
i3 <- bins == 3 & !is.na(bins)
i4 <- bins == 4 & !is.na(bins)
x.p1 <- dens$x[i1]; x.p1 <- c(x.p1,max(x.p1),min(x.p1))
x.p2 <- dens$x[i2]; x.p2 <- c(x.p2,max(x.p2),min(x.p2))
x.p3 <- dens$x[i3]; x.p3 <- c(x.p3,max(x.p3),min(x.p3))
x.p4 <- dens$x[i4]; x.p4 <- c(x.p4,max(x.p4),min(x.p4))
y.p1 <- dens$y[i1]; y.p1 <- c(y.p1,0,0)
y.p2 <- dens$y[i2]; y.p2 <- c(y.p2,0,0)
y.p3 <- dens$y[i3]; y.p3 <- c(y.p3,0,0)
y.p4 <- dens$y[i4]; y.p4 <- c(y.p4,0,0)
plot(dens , type = "n" , axes = F, main = main, xlab = xlab , ylab = ylab)
polygon(x.p1,y.p1 , border = F , col = col)
polygon(x.p2,y.p2 , border = F , col = col)
polygon(x.p3,y.p3 , border = F , col = col)
polygon(x.p4,y.p4 , border = F , col = col)
axis(side = 1 , line = 0 , at = fivenum(x, na.rm = T) , label = c("Minimum","Quartile 1", "Median", "Quartile 3", "Maximum"), lwd = 0, cex.axis = 0.6)
axis(side = 1 , line = 1 , at = fivenum(x, na.rm = T))
axis(side = 1 , line = 1 , at = round(mean(x , na.rm = T) , digits = digits) , tcl = 0.4 , label = F)
axis(side = 1 , line = -1.5 , at = round(mean(x , na.rm = T) , digits = digits) , tick = F , cex.axis = 0.6)
axis(side = 1 , line = -2.0 , at = round(mean(x , na.rm = T) , digits = digits) , label = "Mean" , tick = F , cex.axis = 0.6)
}
Some might find the colored area to much, although IMHO it puts the focus on the fact that one is looking at areas when ploting a density. But then something similar without the color fill. Not a function just a few lines of code to embed and adjust to the script:
… and the source…
plot(density(angio$PE_ALDER , na.rm = T), axes = F, main = "Basic densityplot", xlab = "" , ylab = "")
# Add Quartiles
axis(side = 1 , line = 1 , at = fivenum(angio$PE_ALDER, na.rm = T) , label = c("Minimum","Quartile 1", "Median", "Quartile 3", "Maximum"), lwd = 0, cex.axis = 0.6)
axis(side = 1 , line = 2 , at = fivenum(angio$PE_ALDER, na.rm = T))
abline(v = fivenum(angio$PE_ALDER, na.rm = T)[2:4] , lty = 3)
# Mean
axis(side = 1 , line = 2 , at = round(mean(angio$PE_ALDER , na.rm = T) , digits = 2) , tcl = 0.4 , label = F)
axis(side = 1 , line = -0.5 , at = round(mean(angio$PE_ALDER , na.rm = T) , digits = 2) , tick = F)
axis(side = 1 , line = -1.4 , at = round(mean(angio$PE_ALDER , na.rm = T) , digits = 2) , label = "Mean" , tick = F , cex.axis = 0.6)
abline(v = mean(angio$PE_ALDER , na.rm = T) , lty = 4)
… finally an even more stripped down version without the mean:
which was achieved by:
plot(density(angio$PE_ALDER , na.rm = T), axes = F, main = "Basic densityplot", xlab = "Age")
abline(v = fivenum(angio$PE_ALDER, na.rm = T)[2:4] , lty = 3)
axis(side = 1 , line = -1 , at = fivenum(angio$PE_ALDER, na.rm = T) , label = c("Minimum","Quartile 1", "Median", "Quartile 3", "Maximum"), lwd = 0, cex.axis = 0.6)
axis(side = 1 , at = fivenum(angio$PE_ALDER, na.rm = T))
Rforge 