class: center, middle, inverse, title-slide # Functions in <code>R</code> ### Sam Tyner ### 2018/06/05 --- class: primary # Functions A **function** in `R` takes *input* and returns *output* - Input is given in the form of *arguments* passed to the function - Output is what is *returned* from the function --- class: primary # Function help Help files usually have 6 most useful sections - **Description**: tells you what the function does (may not be super informative) - **Usage**: shows you how to use the function and all the arguments it takes with their defaults<sup>1<sup/> - **Arguments**: describes the arguments used in the function and what type of value each one takes. There must always be a comma `,` between arguments. - **Details**: tells you more about how the function behaves - **Value**: describes what the function's output is - **Examples**: shows you (hopefully working) examples of how to use the function .footnote[1. Note: the default values of a function will be taken as arguments unless you change them. If a function argument has no default, you must provide a value for this argument.] --- class: primary # Your Turn 2.1 1. Create a numeric vector of length 10 called `x` 2. Look at the help file for the function `sample()` (Hint: `?sample`) 3. Create a new vector of length 100 called `samp_x` that contains a sample of size 100 from `x` with replacement. --- class: primary # Base `R` Right now, you "only" have access to base `R` functions. Base `R` is incredibly powerful, and you can do many powerful statistical procedures, such as: - simulation: `rnorm()`, `rbinom()`, `rt()`, and many more - Student's `\(t\)`-test: performs tests for population means (`t.test()`) - Fitting linear models with `lm()` - Fitting generalized linear models with `glm()` - and many more! --- class: primary # Wait, there's more! We can install additional functionality in the form of `R` *packages* An `R` package: - Contains a set of specialized functions - Usually all the functions are related to performing one type of task or data analysis (e.g. data cleaning, analyzing your Fitbit data, analyzing text data, etc.) - Can be installed by running `install.packages("packagename")` where `packagename` is the name of the package. It must be in quotes - Is only recognized by `R` when you `library` it. * Think of it like a library: if the book (here, a function in a package) isn't in the library, `R` can't find it and will return an error. * Access the functions in the package by running `library("packagename")` or `library(packagename)`. Note you can use the quotes or not. (People are lazy so `R` accommodated them in this case.) --- class: primary # Your Turn 2.2 One of the packages we will use is actually many packages in one! It is called `tidyverse`. You can read about it at [www.tidyverse.org](https://www.tidyverse.org/). Install this package and tell `R` you want to use its functions. --- class: primary # Function Errors When a function errors... - Don't panic! - You didn't "break" anything - `R` is dumb: it only does *exactly* what you tell it to do - If you told it to do something it can't read or can't do, it will error. - Trial and error is a constant with `R`. If at first you don't succeed, try and try again! --- class: primary # Common Errors Here are some very common things beginners do that will cause `R` to error: - Forgetting a comma `,` after an argument in a function - Forgetting a parenthesis `)` or having too many parentheses. Each `(` must have a parter `)`. These are called the "open parenthesis" and the "close parenthesis" - Telling `R` to use a function in a package you haven't libraried yet - Using an object in a function that you haven't created yet --- class: primary # Error Messages Some error messages are more informative than others - `Error: object 'my_data' not found` means you are trying to use the `my_data` object but haven't created it yet. - `Error in fun(arg): could not find function "fun"` means you are trying to use a function that you haven't libraried OR your spelled the function wrong - `Error: unexpected numeric constant in "yourcode"` means you probably forgot a comma after an argument in the function If you don't know what an error means, just Google it --- class: primary # Function shortcuts RStudio has some built-in features that make writing code easier: - Autocomplete function name: type the first few letters of a function and hit Tab. Several functions show up and you can arrow through them and hit Enter when you find the one you want. Helpful if you forget how a function is spelled! - Hover box showing function usage: after hitting Enter, a box pops up showing the Usage of the funtion so you know what arguments the function takes - Argument completion: inside the empty `()` hit Tab to see all the function arguments available. --- class: primary # Finding Functions There are lots of different ways to find out what function to use in `R`: - Google: how to (do something) in R + CRAN - [CRAN Task Views](https://cran.r-project.org/web/views/): curated lists of packages online that are all related to some topic e.g. finance, genetics, graphics, and many more - [StackOverflow](https://stackoverflow.com/questions/tagged/r) --- class: primary # Your Turn 2.3 Take some time to think about a statistical analysis you may be interested in doing. This can be anything from text analysis, to statistics in your favorite professional sport, or really any topic you can think of. 1. Try to find an `R` package that does the analysis you're interested in (or something close) 2. Install the `R` package you find 3. Find an example in the help file and try doing it yourself!