ggplot in R

1. create a plot object: 

p= ggplot(diamonds,aes(carat,price,color=cut))

dataset / mappings :aes()

2. add a layer:

a minimal layer is to specify a geom.

e.g. p=p+layer(geom="point)

p+geom_histogram(binwidth=2, fill="steelblue")

function like this: 

geom_xxx(mappings, data,...,geom,position)

...:parameters for the geom or stats, such as bin_width for histogram or bandwidth for a loess smoother

mappings: specified using aes() optional

data: a dataset overrides the default plot dataset. optional

geom or stat: optional

plot  and layer are all object in R, they can be stored as variables. 

bestfit=geom_smooth(method="lm",se=F,color=alpha("red",0.5,size=2)

qplot(sleep_rem,sleep_total,data=msleep)+bestfit

qplot(awake, brainwt, data=msleep,log="y")+bestfit

qplot (bodywt, brainwt,data=msleetp, log="xy")+bestfit

data: must be a dataframe

replace an old dataset using %+%

Aesthetic mappings:

aes(x=weight,y=height,color=sqrt(age))

functions of variables can be used

setting vs. mapping

p=ggplot(mtcars, aes(mpg,wt))

p+geom_point(color="darkblue")

is very different than 

p+geom_point(aes(color=“darkblue"))

in qplot, use color=I("darkblue") to set a value 

Grouping:

you want to separate your data into groups, but render them in the same way. 

p=ggplot（Oxboys, aes(age,height,group=Subject))+geom_line()

different groups on different layers

p+geom_smooth(aes(group=1），method="lm",size=2,se=F)

Override the default grouping

boysbox=ggplot(Oxboys, aes(Occasion, height))+geom_boxplot()

OVERRIDE the default grouping :

boysbox=boysbox+geom_line(aes(group=Subject),color="#3366FF")

ggplot(diamonds,aes(carat))+geom_histogram(aes(y=..density..),binwidth=0.1)

d=ggplot(diamonds, aes(carat))+xlim(0,3)

d+stat_bin(aes(ymax=..count..),binwidth=0.1)

d+stat_bin(aes(size=..density..),binwidth=0.1,geom="point",position="identify")

d+stat_bin(y=1,fill=..count..), binwidth=0.1,geom="tile",position="identify")

plot different data on the same plot:

a very common example is supplementing the data with predictions from a model.

model=lme(height~age,data=Oxboys,random=~1+age|Subject))

oplot=ggplot(Oxboys,aes(age,height,group=Subject)+geom_line()

age_grid=seq(-1,1,length=10)

subjects=unique(Oxboys$Subjects)

preds=expand.grid(age=age_grid,Subject=subjects)

preds$height=predict(model,preds)

oplot+geom_line(data=preds,color="#3366FF",size=0.4)

model2=update(model,height~age+I(age^2))

Oxboys$fitted=predict(model2)

Oxboys$resid2=with(Oxboys,fitted-height)

oplot %+%Oxboys+aes(y=resid2)+geom_smooth(aes(group=1))