Question

I am creating an R script to produce a chart consisting of multiple lines. I have produced a basic chart, and have managed to smooth each of the plotted lines. However, I want to beautify the chart by doing the following:

1. Currently, each line is plotted as a smoothed line. I want to plot the smoothed line using some other line type e.g. dash-dot, dash-dash-dot etc.

2. I want to add a legend to the chart, and display a line (same color as the line it represents in the main chart), in the legend.

3. Specify font size for the main chart title, x and y

Here is what I have so far:

dat  <- read.csv(filename, sep=',')

xvals <- dat$xvals
y1  <- dat$y1
y2  <- dat$y2
y3  <- dat$y3
y4  <- dat$y4
y5  <- dat$y5
y6  <- dat$y6

lo1 <- loess(y1~xvals)
lo2 <- loess(y2~xvals)
lo3 <- loess(y3~xvals)
lo4 <- loess(y4~xvals)
lo5 <- loess(y5~xvals)
lo6 <- loess(y6~xvals)

plot(xvals,y1, xlab='X label', ylab='Y label', type='n')
xl <- seq(min(xvals),max(xvals), (max(xvals) - min(xvals))/1000)
lines(xl, predict(lo1,xl), col='gray', lwd=1)
lines(xl, predict(lo2,xl), col='pink', lwd=1)
lines(xl, predict(lo3,xl), col='red', lwd=1)
lines(xl, predict(lo4,xl), col='cyan', lwd=1)
lines(xl, predict(lo5,xl), col='black', lwd=1)
lines(xl, predict(lo6,xl), col='green', lwd=1)

legend("topright", "(x,y)", pch=1, lty=c(1,1), # gives the legend appropriate symbols (lines)
       , lwd=c(1,1),col=c("blue","red"), inset = .02)

How do I modify the code above to implement the requirements 1 – 3 above?

As an aside, I am aware of ggplot. I think I’ll stick with plot() – I find ggplot far too cryptic, puzzling and infuriating.

Answer 1

Since I had started work on an answer already:

Make up some data:

dat <- data.frame(xvals=1:100,
                  y=matrix(rnorm(600),ncol=6))

Create formulae, apply loess and predict, collapse the results into a matrix:

formulae <- lapply(names(dat)[-1],
                   function(yvar) as.formula(paste(yvar,"xvals",sep="~")))
loessfits <- lapply(formulae,loess,data=dat)

xl <- with(dat,seq(min(xvals),max(xvals), length.out=1000))
pvals <- do.call(cbind,lapply(loessfits,predict,newdata=data.frame(xvals=xl)))

Set parameters (it’s always worth going back and spending some more time reading ?par):

## set box type (open), horizontal tick labels, axis label and title size
par(bty="l",las=1,cex.main=2,cex.lab=2)

Set up colo(u)rs and line types:

cvec <- c("gray","pink","red","cyan","black","green")
lvec <- 1:6

Create the plot:

matplot(xl,pvals,type="l",col=cvec,lty=lvec,
        xlab='X label', ylab='Y label')    
legend("topright", paste0("L",1:6), pch=1, lty=lvec,col=cvec, inset = .02)