functions.Rmd

---
title: "Functions"
author: "Jasmine Lai"
date: "23/10/2019"
output: 
  ioslides_presentation:
    transition: "faster"
---
```{r setup, include=FALSE}
knitr::opts_chunk$set(include = TRUE, warning = FALSE, message = FALSE, comment = "")
```

## Outline
- What are functions
- Writing your own Functions
- Debugging
- Pitfalls

## Why Functions?
- Reduce the amount of copy & pasting required
- Let computers do what they do best: repeatitive tasks
- Code is more readable

## Functions
- You have used Functions in the past like `sum()`, `mean()`, `filter()`
- Under the hood these are powered by lines of R code
- you can see this by going command + click the function name
- does not work for all functions (ie base R)

```{r source}
#filter
#mean
```

## `r emo::ji("cake")` {.smaller}
<div style="float: left; width: 40%;">
### Recipe
Pumpkin Pie `r emo::ji("jack-o-lantern")`

Ingredients:

- Pumpkin
- Egg
- Pie Crust
 
Recipe:

- Preheat the oven to 350 F
- Whisk together pumpkin, egg
- Pour mixture into pie crust
</div>

<div style="float: right; width: 60%;">
### R code
```{r pie function}
Pumpkin_pie <- function(pumpkin, ..., p_crust){
#  1. Preheat the oven to 350 F
#  2. Whisk together pumpkin, egg
#  3. Pour mixture into crust
#  Produces a pi <- automatically `returns`
}
```
</div>

## Basic Structure

```{r structure}
Name_of_function <- function(ingredient_1){
  # Body of your function
  # where your R code goes
  # "Recipe goes here"
}
```

## Demo -  Function for Area 
```{r function demo}
#fill in the template
#variables in your arguments can be referenced in your body
tri_area <- function(height, base){
  height * base / 2 
}
# running it will give you a new function in your environment

tri_area(1, 5)
```

## General rule of thumb:

If you need to copy more than 3 times 
you should start thinking about writing a function

## Tips for writing functions
![MVP](https://stat545.com/img/spotify-howtobuildmvp.jpg)

## General Steps
- Write some working R code
- Put code into the function format
- Test it
- Fix it until it works

## Challenge 1
Write a function that:

- given the width, depth and height
- calculates the volume of a rectangular prisim

```{r challenge1, include = FALSE}
vol_rect <- function(w,d,h){
  w * d * h
}

vol_rect(3,4,5)
```

## Debugging - specialized functions

- `debug()`
- `browser()`
- rstudio "red dot" debugger

## Debugging with Print!
```{r print}
debug <- function(height, base){
  1 + 2
  height * base / 2
}

debug(1,2)

debug2 <- function(height, base){
  print(1 + 2)
  height * base / 2
}

debug2(1,2)
```

## Pitfalls

- TidyEvaluation (we will focus on this)
- Lexical Scoping (look this up later)

## Pitfalls - Tidy Evaluation

Trying this for ggplot2
```{r loadyourlibrary}
library("ggplot2")
```

Using the built in dataset
```{r mtcarsdataset}
head(mtcars)
```

## But this doesn't work
```{r errormtcars, results = 'hide'}
#ggplot(data = mtcars, aes(x = mpg, y = wt)) +
#    geom_point()

multi_graph <- function(cx, cy){
  ggplot(data= mtcars, aes(x = cx, y = cy)) +
    geom_point()
}

#multi_graph(mpg,wt)
```

## Tidyeval to the Rescue
put a {{}} on everything

```{r tidyeval demo, echo=FALSE}
multi_graph_2 <- function(cx, cy){
  ggplot(data= mtcars, 
         aes(x = {{cx}}, y = {{cy}})) +
    geom_point()
}

multi_graph_2(mpg,wt)
```

## Challenge 2
Create a function that:

- will plot a graph of any column in the mtcars dataset
- saves the box plot under any name given
```{r challenge2, echo=FALSE}
custom_gg <- function(column1, column2, name){
  ggplot(data = mtcars, 
         aes(x = {{column1}}, y = {{column2}})) +
    geom_boxplot()
  
  ggsave(name)
}

custom_gg(mpg, cyl, "new_plot.jpg")
```