# Getting help

## Introduction {#console-help}

R has a comprehensive built-in help system. This system is orientated around the base R functions and packages. Every good package comes with a set of __help files__. At a minimum these should provide information about the individual package functions and summaries of the data included with the package. They sometimes give descriptions of how different parts of the package should be used, and if we're lucky, one or more "vignettes" that offer a practical demonstration of how to use a package. Other files are sometimes shipped with packages. For example, these might give an overview of the mathematical or computational theory a package relies on. We will not worry about these in this course.

We may as well get something out of the way early on. The word "help" in the phrase "help file" is a bit of a misnomer. It is probably more accurate to say R has an extensive __documentation__ system. The reason we say this is that the majority of help files are associated with functions, and these kinds of files are designed first and foremost to document how a particular function or group of functions are meant to be used. For example, they describe what kinds of arguments a function can take and what kind of objects it will return to us. Help files are also written in a very precise, painful-to-read manner. They contain a lot of jargon which can be hard for new R users to decipher.

The take-home message is that R help files are aimed more at experienced users than novices. Their primary purpose is to carefully document the different elements of a package, rather than explain how a particular function or the package as whole should be used to achieve a given end. That said, help files often contain useful examples, and many package authors do try to make our life easier by providing functional demonstrations of their package (those "vignettes" we mentioned above are a vehicle for doing this). It's important to try to get to grips with the built in help system. It contains a great deal of useful information which we need to really start using R effectively. The road to enlightenment is bumpy though.

## Browsing the help system

How do we access the help system? Help files are a little like mini web pages, which means we can navigate among them using hyperlinks. This makes it very easy to explore the help system. One way to begin browsing the help system uses the `help.start` function:

```{r, eval=FALSE}

help.start()

```

If we type this now at the Console we should see the __Package Index__ page open up in the __Help__ tab of the bottom right pane in RStudio. This lists all the packages currently installed on a computer. We can view all the help files associated with a package by clicking on the appropriate link. For example, the functions that come with the base installation of R have a help file associated with them---click on the link to the R base package (`base`) to see these. Though we know about a few of these already, there are __a lot__ of functions listed here. R is huge.

The packages that come with the base R installation and those that we install separately from base R have their own set of associated help files. These can be viewed by following the appropriate link on the __Package Index__ page. We will learn how to navigate these in a moment. Take note: it is up to the developer of a package to produce usable help files. Well-designed packages like __dplyr__ and __ggplot2__ have an extensive help system that covers almost everything the package can do. This isn't always the the case though, particularly with new or packages or packages that are not widely used. We will only ever use well-documented packages.

Notice that the help browser has Forward, Back, and Home buttons, just like a normal web browser. If we get lost in the mire of help pages we can always navigate backward until we get back to a familiar page. However, for some reason the Home button does not take us to the same page as `help.start`. Clicking on the home button takes us to a page with three sections:

1. The __Manuals__ section looks like it might be useful for novice users. Unfortunately, it isn't really. Even the "Introduction to R" manual is only helpful for someone with a bit of programming experience because it assumes we understand what terms like "data structure" and "data type" mean. It is worth reading this manual after gaining a bit of experience with R. The others manuals are more or less impenetrable unless the reader already knows quite a bit about computing in general.

2. The __Reference__ section is a little more helpful. The "Packages" link just takes us to the same page opened by `help.start`. From here we can browse the help pages on a package-specific basis. The "Search Engine & Keywords" link takes us to a search engine page (no surprises there). We can use this to search for specific help pages, either by supplying a search term or by navigating through the different keywords. We'll discuss the built-in search engine in the next subsection.

3. The __Miscellaneous Material__ section has a couple of potentially useful links. The "User Manuals" link lists any user manuals supplied by package authors. These tend to be aimed at more experienced users and the packages we will learn to use in this course do not provide them. However, it is worth knowing these exist as they are occasionally useful. The "Frequently Asked Questions" link is definitely worth reviewing at some point, but again, most of the FAQs are a little difficult for novice users to fully understand.

## Searching for help files

After browsing help files via `help.start` for a bit it quickly becomes apparent that this way of searching for help is not very efficient. Quite often we know the name of the function we need to use and all we want to do is open that particular help file. We can do this with the `help` function:

```{r, eval=FALSE}

help(topic = Trig)

```

After we run this line RStudio opens up the help file for the trigonometry topic in the __Help__ tab. This file provides information about the various trigonometric functions such as `sin` or `cos`. We'll learn how to make sense of such help pages in the next subsection. For now, we just want to see how to use `help`.

The `help` function needs a minimum of one argument: the name of the topic or function of interest. When we use it like this the help function searches across packages, looking for a help file whose name gives __an exact match__ to the name we supplied. In this case, we opened the help file associated with the `Trig` topic. Most of the time we use the `help` function to find the help page for a specific function, rather than a general topic. This is fine if we can remember the name of the topic associated with different functions. Most of us cannot. Luckily, the help function will also match help pages by the name of the function(s) they cover:

```{r, eval=FALSE}

help(topic = sin)

```

Here we searched for help on the `sin` function. This is part of the `Trig` topic so `help(topic = sin)` brings up the same page as the `help(topic = Trig)`.

There are several arguments of `help` that we can set to alter its behaviour. We will just consider one of these. By default, the `help` function only searches for files associated with the base functions or with packages that we have loaded in the current session with the `library` function. If we want to search for help on the `mutate` function---part of the `dplyr` package---but we haven't run `library(dplyr)` in the current session this will fail:

```{r}

help(mutate)

```

Instead, we need tell `help` where to look by setting the `package` argument:

```{r, eval=FALSE}

help(mutate, package = dplyr)

```

It's good practise to use `help` every time we're struggling with a particular function. Even very experienced R users regularly forget how to use the odd function and have to dive into the help. It's for this reason that R has a built in shortcut for `help`. This is accessed via `?`. For example, instead of typing `help(topic = sin)` into the Console we can bring up the help page for the `sin` function by using `?` like this:

```{r, eval=FALSE}

?sin

```

This is just a convenient shortcut that does the same thing as `help`. The only difference is that `?` does not allow us to set arguments such as `package`.

## Navigating help files {#nav-help}

Navigating a typical help file is a little daunting at first. These files can be quite long and they contain a lot of jargon. The help files associated with functions -- the most common type -- have a consistent structure though. There are a number of distinct sections, whose order is always the same. Wrestling with a help file is much easier if we at least understand what each section is for. After the title, there are eight sections we need to know about:

1. __Description__ gives us a short overview of what the function is meant to be used for. If the help page covers a family of related functions it gives a collective overview of all the functions. Always read this before diving into the rest of the help file.

2. __Usage__ shows how the function(s) are meant to be used. It lists each member of the family as well as their common arguments. The argument names are listed on their own if they have no default, or in name-value pairs, where the value gives the default used should we choose not to set it ourselves when we call the function.

3. __Arguments__ lists each of the allowed arguments, along with a short description of what they influence. This also tells us what what kind of data we are allowed to use with each argument, along with the allowable values (if this is relevant). Always read this section.

4. __Details__ describes precisely how the function(s) behave, often in painful, jargon-laden detail. It is usually the longest and hardest-to-comprehend section but is worth reading as it flags up common "gotchas". We can sometimes get away with ignoring this section, but when we really want to understand a function we need to wade through it.

5. __Value__ explains kind of data structure or object a function returns to us when it finishes doing whatever it does. It's often possible to guess what this will be from the type of function, but nonetheless, it usually a good idea to check whether our reasoning is correct. If it isn't, we probably don't understand the function yet.

6. __References__ just lists the key reference(s). These are worth digging out if we really need to know the 'hows' and 'whys' of a function. We can often skip this information. The one exception is if the function implements a particular statistical tool. In that case it's sensible to go away and read the literature before trying to use it.

7. __See Also__ gives links to the help pages of related functions. These are usually functions that do something similar to the function of interest or are meant to be used in conjunction with it. We can often learn quite a bit about packages or related functions by following the links in this section.

8. __Examples__ provides one or more examples of how to use the function. These are stand-alone examples, so there's nothing to stop us running them. This is often the most useful section of all. Seeing a function in action is a very good way to cut through the jargon and understand how it works.

## Vignettes and demos {#vignettes}

The Oxford English Dictionary defines a vignette as, "A brief evocative description, account, or episode." The purpose of a package vignette in R is to provide a relatively brief, practical account of one or more of its features. Not all packages come with vignettes, though many of the best thought out packages do. We use the `vignette` function to view all the available vignettes in Rstudio. This will open up a tab that lists each vignette under their associated package name along with a brief description. A package will often have more than one vignette. If we just want to see the vignettes associated with a particular package, we have to set the `package` argument. For example, to see the vignettes associated with __dplyr__ we use:

```{r, eval=FALSE}

vignette(package = "dplyr")

```

Each vignette has a name (the "topic") and is available either as a PDF or HTML file (or both). We can view a particular vignette by passing the `vignette` function the `package` and `topic` arguments. For example, to view the "data_frames" vignette in the __`dplyr`__ package we would use:

```{r, eval=FALSE}

vignette(topic = "data_frames", package = "dplyr")

```

The `vignette` function is fine, though it is usually more convenient to browse the list of vignettes inside a web browser. This allows us to open a particular vignette directly by clicking on its link, rather than working at the Console. We can use the `browseVignettes` function to do this:

```{r, eval=FALSE}

browseVignettes()

```

This will open a page in our browser showing the vignettes we can view. As one should expect by now, we can narrow our options to a specific package by setting the `package` argument.

In addition to vignettes, some packages also include one or more 'demos' (demonstrations). Demos are a little like vignettes, but instead of just opening a file for us to read, the demo function can actually runs a demonstration R scripts. We use the `demo` function (without any arguments) to list the available demos:

```{r, eval=FALSE}

demo()

```

When we use the `demo` function like this it only lists the demos associated with packages that have been loaded in the current session (via `library`). If we want to see all the demos we can run we need to use the somewhat cryptic `demo(package = .packages(all.available = TRUE))`.

In order to actually run a demo we use the `demo` function, setting the `topic` and `package` arguments. For example, to run the "colors" demo in the __grDevices__ package we would use:

```{r, eval=FALSE}

demo(colors, package = "grDevices", ask = FALSE)

```

This particular demo shows off some of the pre-defined colours we might use to customise the appearance of a plot. We've suppressed the output though because so much is produced.