R for reproducible scientific analysis

Reference

Introduction to R and RStudio

• Use the escape key to cancel incomplete commands or running code (Ctrl+C) if you're using R from the shell.
• Basic arithmetic operations follow standard order of precedence:
• Brackets: (, )
• Exponents: ^ or **
• Divide: /
• Multiply: *
• Add: +
• Subtract: -
• Scientific notation is available, e.g: 2e-3
• Anything to the right of a # is a comment, R will ignore this!
• Functions are denoted by function_name(). Expressions inside the brackets are evaluated before being passed to the function, and functions can be nested.
• Mathematical functions: exp, sin, log, log10, log2 etc.
• Comparison operators: <, <=, >, >=, ==, !=
• Use all.equal to compare numbers!
• <- is the assignment operator. Anything to the right is evaluate, then stored in a variable named to the left.
• ls lists all variables and functions you've created
• rm can be used to remove them
• When assigning values to function arguments, you must use =.

Project management with RStudio

• To create a new project, go to File -> New Project
• Install the packrat package to create self-contained projects
• install.packages to install packages from CRAN
• library to load a package into R
• packrat::status to check whether all packages referenced in your scripts have been installed.

Reading data

• read.table to read in data in a regular structure
• sep argument to specify the separator
  • "," for comma separated
  • "" for tab separated
• Other arguments:
  • header=TRUE if there is a header row

Seeking help

• ? or help() to seek help for a function.
• ?? to search for a function.
• Wrap special operators in quotes when searching for help: help("+").
• CRAN Task Views.
• stackoverflow.

Data structures

Basic data structures in R:

• atomic ?vector (can only contain one type)
• ?list (containers for other objects)
• ?data.frame two dimensional objects whose columns can contain different types of data
• ?matrix two dimensional objects that can contain only one type of data.
• ?factor vectors that contain predefined categorical data.
• ?array multi-dimensional objects that can only contain one type of data

Remember that matrices are really atomic vectors underneath the hood, and that data.frames are really lists underneath the hood (this explains some of the weirder behaviour of R).

Data types:

• ?numeric real (decimal) numbers
• ?integer whole numbers only
• ?character text
• ?complex complex numbers
• ?logical TRUE or FALSE values

Special types:

• ?NA missing values
• ?NaN "not a number" for undefined values (e.g. 0/0).
• ?Inf, -Inf infinity.
• ?NULL a data structure that doesn't exist

NA can occur in any atomic vector. NaN, and Inf can only occur in complex, integer or numeric type vectors. Atomic vectors are the building blocks for all other data structures. A NULL value will occur in place of an entire data structure (but can occur as list elements).

Useful functions for querying data structures:

• ?str structure, prints out a summary of the whole data structure
• ?typeof tells you the type inside an atomic vector
• ?class what is the data structure?
• ?head print the first n elements (rows for two-dimensional objects)
• ?tail print the last n elements (rows for two-dimensional objects)
• ?rownames, ?colnames, ?dimnames retrieve or modify the row names and column names of an object.
• ?names retrieve or modify the names of an atomic vector or list (or columns of a data.frame).
• ?length get the number of elements in an atomic vector
• ?nrow, ?ncol, ?dim get the dimensions of a n-dimensional object (Won't work on atomic vectors or lists).

Data subsetting

• Elements can be accessed by:
• Index
• Name
• : to generate a sequence of numbers to extract slices
• [ single square brackets:
• extract single elements or subset: - vectors
• extract single elements of a list
• extract columns from a data.frame
• [ with two arguments to:
• extract rows and/or columns of
  • matrices
  • data.frames
• [[ double square brackets to subset lists
• $ to access columns or list elements by name
• negative indices skip elements

Writing data

• write.table to write out objects in regular format
• set quote=FALSE so that text isn't wrapped in " marks

Vectorisation

• Most functions and operations apply to each element of a vector
• * applies element-wise to matrices
• %*% for true matrix multiplication
• any() will return TRUE if any element of a vector is TRUE
• all() will return TRUE if all elements of a vector are TRUE

Control flow

• Use if condition to start a conditional statement, else if condition to provide additional tests, and else to provide a default
• The bodies of the branches of conditional statements must be indented.
• Use == to test for equality.
• X && Y is only true if both X and Y are TRUE.
• X || Y is true if either X or Y, or both, are TRUE.
• Zero is considered FALSE; all other numbers are considered TRUE
• Nest loops to operate on multi-dimensional data.

Functions

• Put code whose parameters change frequently in a function, then call it with different parameter values to customize its behavior.
• The last line of a function is returned, or you can use return explictly
• Any code written in the body of the function is isolated to the function when it is called.
• Document Why, then What, then lastly How (if the code isn't self explanatory)

Split-apply-combine

• Use the xxply family of functions to apply functions to groups within some data.
• the first letter, array , data.frame or list corresponds to the input data
• the second letter denotes the output data structure
• Anonymous functions (those not assigned a name) are used inside the plyr family of functions on groups within data.

GGplot2

• figures can be created with the grammar of graphics:
• ggplot to create the base figure
• aesthetics specify the data axes, shape, color, and data size
• geometry functions specify the type of plot, e.g. point, line, density, box
• geometry functions also add statistical transforms, e.g. geom_smooth
• scale functions change the mapping from data to aesthetics
• facet functions stratify the figure into panels
• aesthetics apply to individual layers, or can be set for the whole plot inside ggplot.
• theme functions change the overall look of the plot
• order of layers matters!
• ggsave to save a figure.

Defensive Programming

• Program defensively, i.e., assume that errors are going to arise, and write code to detect them when they do.
• Write tests before writing code in order to help determine exactly what that code is supposed to do.
• Know what code is supposed to do before trying to debug it.
• Make it fail every time.
• Make it fail fast.
• Change one thing at a time, and for a reason.
• Keep track of what you've done.
• Be humble