R Fundamentals

Submit Attendance: link

Files for today: example | vectors | matrices | lists

Object Types

  • Variables
    • Variables are named objects; objects result from the evaluation of an expresssion
    • Names can be assigned with an equals sign (=) or the assignment operator (<-, alt+minus shortcut)
    • For example, with x <- 2+2, the expression 2+2 evaluates to a length-one vector holding value 4, to which we have bound the name x
  • Many ways to check the “type” of data in an R object
    • My preferred method is to assess the structure of an object using str()
    • Additional options include mode() / typeof() / class() among others
  • You can check data types or convert between data types
    • For example, using is.character() to check or as.character() to convert
  • Special “values” in R include:
    • TRUE and FALSE logical values (coerced to 0 and 1 when used in mathematical operations)
    • NA indicates a placeholder for a missing value
    • Inf and NaN result from, eg, dividing by zero
    • NULL indicates a placeholder for a missing vector object
  • It is helpful to think that everything in R is either an object or a function (and functions themselves are objects)
    • The expression 2+2 is actually a call to the +() function, as in `+`(2,2)

Vectors

  • R automatically performs vector recycling when operating on multiple vectors of different lengths
    • This makes a lot of sense when one of the vectors has length-one
    • It can lead to nice “tricks” in very special circumstances
    • However, in general for vectors longer than length-one, it is unusual behavior
  • Vector can be subset using square brackets [] by:
    • Position number, or a vector of position numbers – no zero-based indexing like Python or C++
    • Logical vectors, often of the same length as the vector to be subset
    • Element names, if the elements of the vector are named
  • Logical tests are extremely common
    • Use double-equals == to test for equality
    • Use != to indicate not-equal-to
  • R heavily utilizes vectorized functions
    • For example, if x=c(4,9,16) then sqrt(x) returns (2,3,4)

Matrices

  • Matrices are vectors with a dimension attribute
  • Matrix math, ie Linear Algebra, is common in statistical applications
    • Standard operators (+, -, *, /) are element-wise
    • Matrix multiplcation uses the %*% operator
    • Transpose via t()
    • Invert via solve() or a Choleski decomposition
  • Subset a matrix with square brackets []
    • Uses form [row index, column index]
    • Blank index indicates “all”
  • You can “sweep” a function over the rows or columns
    • Simple functions are built-in, such as RowSums() and ColMeans()
    • Alternatively, use apply()

Lists

  • Lists are generalized vectors
    • Each element of a vector holds a value, each element of a list “holds” an R object
    • You can nest lists; ie a list element could be another list
    • Most algorithmic functions in R return lists, eg, lm(), kmeans(), prcomp()
  • List subsetting / extracting elements
    • Single brackets [] return a list object
    • Double brackets [[]] or dollar sign $ extract a list element
  • It is common to apply a function to each object in a list
    • lapply() returns a list of results, sapply() returns a (simplified) vector