Storing results

It is almost always better to create an object to store your results first, rather than growing the object as you go.

# Good
res = rep(NA,10)
for(x in 1:10)
{
  res[x] = x^2
}
res

##  [1]   1   4   9  16  25  36  49  64  81 100

# Bad
res = c()
for (x in 1:10)
{
  res = c(res,x^2)
}
res

##  [1]   1   4   9  16  25  36  49  64  81 100

Back to for loops

Often we want to use a loop across the indexes of an object and not the elements themselves. There are several useful functions to help you do this: :, seq, seq_along, seq_len, etc.

l = list(1:3, LETTERS[1:7], c(TRUE,FALSE))
res = rep(NA, length(l))

for(x in seq_along(l))
{
  res[x] = length(l[[x]])
}

res

## [1] 3 7 2

1:length(l)

## [1] 1 2 3

seq_along(l)

## [1] 1 2 3

seq_len(length(l))

## [1] 1 2 3

Some lessons learned

• Everything we’ve shown so far can also be done using
  • subsetting ([]) or
  • functional approaches (*apply)

• There are almost always multiple possible approaches,
  • the best initial solution is the one you can get working the quickest
  • once something is working you can worry about making it faster / more efficient.

Programmers waste enormous amounts of time thinking about, or worrying about, the speed of noncritical parts of their programs, and these attempts at efficiency actually have a strong negative impact when debugging and maintenance are considered. We should forget about small efficiencies, say about 97% of the time: premature optimization is the root of all evil. Yet we should not pass up our opportunities in that critical 3%.

formals(gcd)

## $loc1
## 
## 
## $loc2

body(gcd)

## {
##     deg2rad = function(deg) return(deg * pi/180)
##     lat1 = deg2rad(loc1[1])
##     lat2 = deg2rad(loc2[1])
##     long1 = deg2rad(loc1[2])
##     long2 = deg2rad(loc2[2])
##     R = 6371
##     d = acos(sin(lat1) * sin(lat2) + cos(lat1) * cos(lat2) * 
##         cos(long2 - long1)) * R
##     return(d)
## }

Argument names

When defining a function we are also implicitly defining names for the arguments, when calling the function we can use these names to

f = function(x,y,z) paste0("x=",x," y=",y," z=",z)

f(1,2,3)

## [1] "x=1 y=2 z=3"

f(z=1,x=2,y=3)

## [1] "x=2 y=3 z=1"

f(y=2,1,3)

## [1] "x=1 y=2 z=3"

f(y=2,1,x=3)

## [1] "x=3 y=2 z=1"

f(1,2,3,m=1)

## Error in f(1, 2, 3, m = 1): unused argument (m = 1)

Scoping

R has generous scoping rules, if it can’t find a variable in the functions body’s scope, it will look for it in the next higher scope, and so on.

y = 1
f = function(x)
{
  x+y
}
f(3)

## [1] 4

g = function(x)
{
  y=2
  x+y
}
g(3)

## [1] 5

Additionally, variables defined within a scope only persist for the duration of that scope, and do not overwrite variables at a higher scopes.

x = 1
y = 1
z = 1
f = function()
{
    y = 2
    g = function()
    {
      z = 3
      return(x + y + z)
    }
    return(g())
}
f()

## [1] 6

c(x,y,z)

## [1] 1 1 1

Getting Help

Prefixing any function name with a ? will open the related help file for that function.

?`+`
?sum

For functions not in the base package, you can also see their implementation by entering the function name without parentheses (or using body function).

lm

## function (formula, data, subset, weights, na.action, method = "qr", 
##     model = TRUE, x = FALSE, y = FALSE, qr = TRUE, singular.ok = TRUE, 
##     contrasts = NULL, offset, ...) 
## {
##     ret.x <- x
##     ret.y <- y
##     cl <- match.call()
##     mf <- match.call(expand.dots = FALSE)
##     m <- match(c("formula", "data", "subset", "weights", "na.action", 
##         "offset"), names(mf), 0L)
##     mf <- mf[c(1L, m)]
##     mf$drop.unused.levels <- TRUE
##     mf[[1L]] <- quote(stats::model.frame)
##     mf <- eval(mf, parent.frame())
##     if (method == "model.frame") 
##         return(mf)
##     else if (method != "qr") 
##         warning(gettextf("method = '%s' is not supported. Using 'qr'", 
##             method), domain = NA)
##     mt <- attr(mf, "terms")
##     y <- model.response(mf, "numeric")
##     w <- as.vector(model.weights(mf))
##     if (!is.null(w) && !is.numeric(w)) 
##         stop("'weights' must be a numeric vector")
##     offset <- as.vector(model.offset(mf))
##     if (!is.null(offset)) {
##         if (length(offset) != NROW(y)) 
##             stop(gettextf("number of offsets is %d, should equal %d (number of observations)", 
##                 length(offset), NROW(y)), domain = NA)
##     }
##     if (is.empty.model(mt)) {
##         x <- NULL
##         z <- list(coefficients = if (is.matrix(y)) matrix(, 0, 
##             3) else numeric(), residuals = y, fitted.values = 0 * 
##             y, weights = w, rank = 0L, df.residual = if (!is.null(w)) sum(w != 
##             0) else if (is.matrix(y)) nrow(y) else length(y))
##         if (!is.null(offset)) {
##             z$fitted.values <- offset
##             z$residuals <- y - offset
##         }
##     }
##     else {
##         x <- model.matrix(mt, mf, contrasts)
##         z <- if (is.null(w)) 
##             lm.fit(x, y, offset = offset, singular.ok = singular.ok, 
##                 ...)
##         else lm.wfit(x, y, w, offset = offset, singular.ok = singular.ok, 
##             ...)
##     }
##     class(z) <- c(if (is.matrix(y)) "mlm", "lm")
##     z$na.action <- attr(mf, "na.action")
##     z$offset <- offset
##     z$contrasts <- attr(x, "contrasts")
##     z$xlevels <- .getXlevels(mt, mf)
##     z$call <- cl
##     z$terms <- mt
##     if (model) 
##         z$model <- mf
##     if (ret.x) 
##         z$x <- x
##     if (ret.y) 
##         z$y <- y
##     if (!qr) 
##         z$qr <- NULL
##     z
## }
## <bytecode: 0x7ffac4613b08>
## <environment: namespace:stats>

Replacement functions

We can also define functions that allow for ‘inplace’ modification like attr or names.

`last<-` = function(x, value)
{
  x[length(x)] = value
  x
}

x = 1:10
last(x) = 5L
x

##  [1] 1 2 3 4 5 6 7 8 9 5

last(1)

## Error in eval(expr, envir, enclos): could not find function "last"