ggplot2 Version of Figures in “25 Recipes for Getting Started with R”

August 16, 2011

(This article was first published on YGC » R, and kindly contributed to R-bloggers)

In order to provide an option to compare graphs produced by basic internal plot function and ggplot2, I recreated the figures in the book, 25 Recipes for Getting Started with R, with ggplot2.

The code used to create the images is in separate paragraphs, allowing easy comparison.

1.16 Creating a Scatter Plot

plot(cars)

ggplot(cars,aes(speed,dist))+geom_point()

1.17 Creating a Bar Chart

heights <- tapply(airquality$Temp, airquality$Month, mean)
par(mfrow=c(1,2))
barplot(heights)
barplot(heights,
        main="Mean Temp. by Month",
        names.arg=c("May", "Jun", "Jul", "Aug", "Sep"),
        ylab="Temp (deg. F)")

require(gridExtra)
heights=ddply(airquality,.(Month), mean)
heights$Month=as.character(heights$Month)
p1 <- ggplot(heights, aes(x=Month,weight=Temp))+
    geom_bar()
p2 <- ggplot(heights, aes(x=factor(heights$Month,
                          labels=c("May", "Jun", "Jul", "Aug", "Sep")),
                          weight=Temp))+
    geom_bar()+
    opts(title="Mean Temp. By Month") +
    xlab("") +
    ylab("Temp (deg. F)")

grid.arrange(p1,p2, ncol=2)

1.18 Creating a Box Plot

y <- c(-5, rnorm(100), 5)
boxplot(y)

ggplot()+geom_boxplot(aes(x=factor(1),y=y))+xlab("")+ylab("")

1.19 Creating a Histogram

data(Cars93, package="MASS")
par(mfrow=c(1,2))
hist(Cars93$MPG.city)
hist(Cars93$MPG.city, 20)

p <- ggplot(Cars93, aes(MPG.city))
p1 <- p + geom_histogram(binwidth=diff(range(Cars93$MPG.city))/5)
p2 <- p + geom_histogram(binwidth=diff(range(Cars93$MPG.city))/20)
grid.arrange(p1,p2, ncol=2)

1.23 Diagnosing a Linear Regression

data(iris)
m = lm( Sepal.Length ~ Sepal.Width, data=iris)
par(mfrow=c(2,2))
plot(m)

r <- residuals(m)
yh <- predict(m)
scatterplot <- function(x,y, title="", xlab="", ylab="") {
	d <- data.frame(x=x,y=y)
	p <- ggplot(d, aes(x=x,y=y)) + geom_point() + opts(title=title) + xlab(xlab) + ylab(ylab)
	return(p)
}

p1 <- scatterplot(yh,r,
                  title="Residuals vs Fitted",
                  xlab="Fitted values",
                  ylab="Residuals")
p1 <- p1 +geom_hline(yintercept=0)+geom_smooth()

s <- sqrt(deviance(m)/df.residual(m))
rs <- r/s

qqplot <- function(y,
                   distribution=qnorm,
                   title="Normal Q-Q",
                   xlab="Theretical Quantiles",
                   ylab="Sample Quantiles") {
    require(ggplot2)
    x <- distribution(ppoints(y))
    d <- data.frame(x=x, y=sort(y))
    p <- ggplot(d, aes(x=x, y=y)) +
        geom_point() +
            geom_line(aes(x=x, y=x)) +
                opts(title=title) +
                    xlab(xlab) +
                        ylab(ylab)
    return(p)
}

p2 <- qqplot(rs, ylab="Standardized residuals")

sqrt.rs <- sqrt(abs(rs))
p3 <- scatterplot(yh,sqrt.rs,
                  title="Scale-Location",
                  xlab="Fitted values",
                  ylab=expression(sqrt("Standardized residuals")))
p3 <- p3 + geom_smooth()

hii <- lm.influence(m, do.coef = FALSE)$hat
p4 <- scatterplot(hii,rs)
p4 <- p4+
    geom_hline(yintercept=0)+
    geom_smooth() +
    geom_text(aes(x=min(hii)+diff(range(hii))*0.3,
                  y=min(rs)+diff(range(rs))*0.04,
                  label="--   Cook's distance", size=3))+
    opts(legend.position="none")

grid.arrange(p1,p2,p3,p4, ncol=2)