Decision tree model is fitted to toy data on homes in New York and San Francisco (NY, SF). Along the way overfitting is illustrated. The optimally fitted tree is used to classify some test data as either NY or SF.

Fixed and mixed models are fitted to the sleepstudy dataset to investigate how human reaction slows down with sleep deprivation.

1. general task: evaluate arbitrary elements of a sequence of expressions
2. specific task: produce normal Q–Q plots for a sequence of distributions
3. two implementation strategies are discussed: one based on switch and another on substitute
4. using substitute demands more abstract programming but affords more flexibility

Iterative processes are characterized by successive steps forming a sequence of states captured by state variables. In each step the state variables are updated based on themselves and some rule. Iteration is often implemented using control-flow constructs like the for or while loop (see ?Control for help).

Graphic packages like lattice and ggplot2 work best when data are presented in long format as opposed to wide format into which data are typically imported and in which it is usually convenient to perform calculations. For example morley is a data frame in long format, which allows functions like lattice::xyplot to receive simple formulas and yet produce a complex trellis display.