Lecture 7 Understanding functions.Rmd

---
title: "Lecture 7 Understanding functions"
author: "David A McAllister"
date: "Monday 31st October, 2016"
output:
  slidy_presentation: default
---

## Objective of section
* To describe parts of a function
* To describe how R matches arguments
* To understand where R looks for values
* To understand how functions (don't) modify the workspace
* To identify side-effects

# See Hadley Wickham again

## 
### Everything that exists is an object. 
### Everything that happens is a function call
### John Chambers

## Function components

All R functions have three parts

* the `body()`, the code inside

* the `formals()`, the list of arguments

* the `environment()`, the "map" of the location of the function's variables

```{r, eval = FALSE}
f <- function(x) x^2
```

## Primitive functions
Primitive functions, like `sum`, call C code directly and contain no R code.
Therefore their `formals()`, `body()` and `environment()` are all `NULL`

## Arguments

* Arguments to functions are matched by R on
    * position, eg `lm (y ~ x, mydata)` 
    * name, eg `lm ( formula = y ~ x, data = mydata)`
    * partial name, eg `lm (form = y ~ x, dat = mydata)`
* name trumps position
    * Use position for 1st or 2nd arguments only `f (,3,,5)`
* Default values are often used ` mean(x, trim = 0, na.rm = FALSE, ...) `
* Defaults can be other arguments `factor (x = character(), levels, labels = levels)`

## An aside about readable code
```{r}
args(glm)
model1 <- glm (formula = "speed ~ dist", family = gaussian, data = cars) # overkill
model1 <- glm (speed ~ dist, data = cars) # about right
model1 <- glm (speed ~ dist, , cars) # awful
```
Use argument name for anything past first or second argument

## Lexical scoping

* Inside function first
* then outside (bad coding)
```{r}
x <- 1
y <- 2
MyFunction <- function (x){
  x * y
}

MyFunction (10)
```

## Only copies modified
```{r}
x <- 1
y <- 2
MyFunction <- function (x,y){
  x <- x+10
  y <- y + 100
  c(x,y)
}
MyFunction (5,15)
c(x,y)
```

## Only returns a single object
### This is why lists are so common in functions

## Except ..., Side effects
``` {r, eval = FALSE}
plot(x)
write.csv (x, file = "myout.csv")
setwd("c:/Users/David/myfolder")
# Or if a function uses superassignment
<<- # considered bad practice
```

## Replacement and infix
```{r}
# Replacement
names(cars)
names (cars) <- c("velocity", "displacement")
names(cars)

# Infix
`+`(1,2)

```
Can you think of any other replacement functions

## A word on quotes

- Generally quotes are needed for variables as these are vectors within dataframes
```{r, eval = FALSE}
newdata <- merge (df1, df2, by = "linkID")
```

- for interactive use, as a convenience some functions accept both
- involves complex programming
  + don't try in your own functions
  + careful about doing it in others functions, might not always work