ggplot2 examples

ggplot_examples.Rmd

---
title: "GrammaR of graphics using `ggplot2` in R"
author: "John Stanton-Geddes"
date: "December 8, 2014"
output: html_document
---

The `ggplot2` package, deveoped by Hadley Wickham, is the most downloaded R package of [all time](http://www.rdocumentation.org/) and one of the standards for publication-ready figures. Fron the [ggplot website](http://ggplot2.org/)

> ggplot2 is a plotting system for R, based on the grammar of graphics, which tries to take the good parts of base and lattice graphics and none of the bad parts. It takes care of many of the fiddly details that make plotting a hassle (like drawing legends) as well as providing a powerful model of graphics that makes it easy to produce complex multi-layered graphics.)

That said, the syntax for generating plots is distinct from that of base R and can take some practice, constant checking of the [documentation](http://docs.ggplot2.org/current/) and searching on [stack overflow](http://stackoverflow.com/questions/tagged/ggplot2) to become familiar with.

Here, I show a few examples.

## Bar chart

From the `ggplot` documentation.

```{r}
library(ggplot2)
library(gridExtra)
library(plyr)

summary(diamonds)

# define plot
c <- ggplot(diamonds, aes(clarity))

# by default, uses stat="bin", which gives the count in each category
c + geom_bar()
```

## Histograms

```{r}
h <- ggplot(diamonds, aes(carat)) 
h + geom_histogram()
h + geom_histogram(binwidth = .1)
```

## Faceting

```{r}
# stacked barplot
ggplot(diamonds, aes(clarity, fill=cut)) + geom_bar()

# facets
ggplot(diamonds, aes(clarity)) + geom_bar() +
  facet_wrap(~ cut)
```

## Boxplots

```{r, echo=FALSE}
d <- ggplot(diamonds, aes(x = color, y = log(price))) + geom_boxplot()
d
```


## Line plots

```{r}
# summarise diamond price by color and carat
ddiamonds <- ddply(diamonds, .(color, carat), summarise, mprice = mean(price))
str(ddiamonds)

# plot
e <- ggplot(ddiamonds, aes(x = carat, y = log(mprice), group = color)) +
  geom_line()
e

# color lines
e + geom_line(aes(colour = color))
```

Lines of the actual data are nice, but smooth lines are better.

```{r}
# smooth
e + geom_smooth(stat = "smooth")
e + stat_smooth(method = "loess")

# lm
e + stat_smooth(method = "lm")
e + stat_smooth(method = "lm", formula = y ~ (poly(x, 3)))
```


Plot multiple time series on a single graph.

```{r}
y2005 <- runif(300, 20, 120)
y2010 <- y2005 * runif(300, -1.05, 1.5)
group <- rep(LETTERS[1:3], each = 100)

df <- data.frame(id = seq_along(group), group, y2005, y2010)
library(reshape2) # for melt
dfm <- melt(df, id.var = c("id", "group"))
ggplot(dfm, aes(variable, value, group = id, colour = group)) +
  geom_path(alpha = 0.5)
# note the alpha parameter that specifies line intensity
```

Here's an example of using ggplot in a Shiny interactive webapp: [https://johnsg.shinyapps.io/ApRxN-shinyapp/]

## qplot

If you're already familiar with base R graphic syntax, you can use `qplot` for most ggplots. For example:

```{r}
qplot(color, log(price), geom = "boxplot", data = diamonds)
```

For full functionality, I recommend just getting used to the `ggplot` syntax. 

## Arrange plots

Even after faceting plots, sometimes you want to group multiple plots into a single figure. In base R, you could do this using combination of `par(mfrow = c(x,x))`. For ggplots, you use the `grid.arrange()` function from the [gridExtra](http://cran.r-project.org/web/packages/gridExtra/index.html) package.

```{r}
e <- ggplot(diamonds, aes(x = color)) + geom_bar()
grid.arrange(d, e, ncol = 1)
```

Note that the [grid package](http://www.inside-r.org/r-doc/grid/viewport) provides a more extensible way to do this using `viewports`, though not quite as simply as `grid.arrange`. 


## Other random things

Violin plot - show the probability desnity at different values. 

```{r}
v <- ggplot(mtcars, aes(factor(cyl), mpg))

v + geom_violin()

# gitter
v + geom_violin() + geom_jitter(height = 0)
```

Rug plot

```{r}
r <- ggplot(mtcars, aes(x=wt, y=mpg))
r + geom_point()
r + geom_point() + geom_rug()
```


## ggvis

The cutting-edge of ggplot is a new package called [ggvis](http://ggvis.rstudio.com/) that allows for interactive plots. Here are a few examples:

```{r}
library(ggvis)
library(dplyr)

# points
mtcars %>% ggvis(~wt, ~mpg) %>% layer_points()


# smooth lines
mtcars %>% 
  ggvis(~wt, ~mpg) %>%
  layer_points() %>%
  layer_smooths()

# grouping
mtcars %>% 
  ggvis(~wt, ~mpg) %>% 
  layer_points(fill = ~factor(cyl))
```


Interactive examples

```{r}
mtcars %>%
  ggvis(~wt, ~mpg) %>%
  layer_smooths(span = input_slider(0.5, 1, value = 1)) %>%
  layer_points(size := input_slider(100, 1000, value = 100))
```