2022-08-21
Learning R can be hard. R objects can be … how may we put it … intricate? Not only is there is a large variety of objects. That is common for well established programming languages. The issue is rather that sometimes things may look similar but have subtle differences. All in all handling R objects can be quite difficult for a beginner. For newbies it can be quite confusing, even intimidating. I know it was for me.
The “huh?” package wants to describe objects to make it easier to understand what they are, what they contain and how to handle them. My success measure: How many times do you open the manual? If “huh!?” reduces the number of occasions you start browsing through pages of the manual or your favourite R textbook, in that case this package is a huge success.
It basically offers only two functions:
huh()how()
… because “Huh!?” must be easy to learn. You do not need to learn many
functions and arguments by heart. You simply ask huh(myObject) or
how(myObject) in the console and get your result.
huh. huh() describes an object. It gives you the basic type,
the class, any attributes including the number of dimensions,
and on what paradigm the object is based on. The paradigm tells you
whether the object is based on S3, S4, reference or R6 classes.
Example.
library(huh)
# integer vector
huh(1:3)
#> name : 1:3
#> type : integer
#> class : integer
#> dimensions: 1
#> paradigm : implicit, object
# integer matrix
xy <- matrix(1:4, 2)
huh(xy)
#> name : xy
#> type : integer
#> class : matrix, array
#> dimensions: 2
#> paradigm : implicit, object
# function (i.e. a closure to be exact)
huh(mean)
#> name : mean
#> type : closure
#> class : function
#> paradigm: S3, generichow. how() tells you how you can access the contents of an object,
which subsetting techniques work, and what the resulting type would be.
Also for several objects how() provides further information that is
useful when you use the objects.
The output groups subsetting techniques by output type so that it is
quite easy what the output of a subsetting operation looks like. A
matrix returns an atomic vector when the subset is 1-dimensional unless
you add the argument drop=FALSE. But since drop=TRUE is the default
how() shows you this:
#> vector
#> y[c(...)], y[c(...), ...], y[[...]]The first line is the output type. The second line shows ways to subset
the matrix that will return this type. ... means that you can only use
a single number or string as index; c(...) allows using a vector of
indices.
Example.
library(huh)
# atomic vector
x <- 1:3
how(x)
#> Subsetting 'x'
#>
#> integer
#> x[c(...)], x[[...]]
# 2-dimensional access
y <- matrix(1:4, 2)
how(y)
#> Subsetting 'y'
#>
#> integer
#> y[c(...)], y[c(...), ...], y[[...]]
#> matrix
#> y[c(...), c(...)]
#> The result is coerced to the lowest possible dimension. I.e. ...
#> * 1-dim results yield atomic vectors,
#> * 2-dim results class 'matrix'
#> Add the argument `drop=FALSE` to keep the class
# function (...well, a closure to be exact)
how(mean) # not subsettable
#> mean is not subsettableThat is all you need to know.
You can install the development version of huh from GitHub with:
# install.packages("devtools")
devtools::install_github("SigurdJanson/huh")The two big steps that are still missing.
- Improved formatting to enhance readability and bring attention to the essential details of the output.
- Enhanced ease of use of the output. It should be possible to just copy/paste output to use it.
- Finish an accompanying vignette about this package and types in R, in general.
Take a look at the issues of the repository if you want to know details.
If you find anything suspicious or you have an idea, just let me know. I am just a regular guy, not an R wizard and I am just trying to be a bit helpful, here. So don’t hesitate if you notice something.
Norberg, R. (2012). Classes and Objects in R. R Bloggers. last access 2022-08-20
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent