R in Linguistic Analysis. Week 2 Wassink Autumn 2012

Transcription

1 R in Linguistic Analysis Week 2 Wassink Autumn 2012

2 Today R fundamentals The anatomy of an R help file but first... How did you go about learning the R functions in the reading? More help learning functions

3 Reading Qs [MO] Step-by-step examples would be helpful for functions sapply(), substr() [NK] Are factors vectors? [NK] I was also wondering about the single vs. double bracket subsetting notation as used by data frames. While x[1] gives the first row of the data frame as a data frame type, and x[[1]] gives the first row of the data frame as its original type, x[1,1] gives the specified cell not as a data frame, but as the original type of that cell. It appears, then, that this form of subsetting is an exception to the rule of what single vs. double brackets mean. [ABW] prop.table() function: I don t see %s that add up to 100. How do I read the output? Ah. 1.0 represents 100%. The text should say proportions, not percentages. [JR] My question doesn't have so much to do with the chapter we read as it does with the bigger picture: what advantages does R have over other languages (e.g. Python)? My sense of it (knowing little of Python, and even less of R) is that R's strengths lie mainly in how it can deal with data structures, and perhaps vectors as Gries suggests. Working with data aside, it seems to me that Python can do everything else we've been doing in chapter 3. What other sorts of things is R commonly used for?

4 The R Interface Command history Quartz Window Editor Window The quartz device window (for tables and graphics) The R Console (for instructions), with or without the The R command history R Data Editor Window (for Excel-type view of working dataset) *note: can type at the command line to access this: > edit(name)

5 Help! getting help Windows: Mac: either: > help.start() >?help >?sqrt Help:HTML help Help:R for MAC OSX FAQs to get to R s online help website if you don t know the function you want if you do.

6 R primer R is an Interpreted language: a programming language in which commands are 'indirectly' executed ("interpreted") by an interpreter program, which executes source code step-by-step, rather than translating it into machine code. This can be contrasted with a compiled language which is converted into machine code and then 'directly' executed by the host central processing unit. (Python, PERL, MATLAB are interpreted; Commands typically consist of: functions (instruction to do something) arguments (target of the operation, how to apply itself) assignment operator <- user-assigned variable or data structure ex. x<-c(1:10) ex. q() Null argument

7 cont. Variables: holders that store values, allow for abstraction Variable names can be any combination of numbers, letters, and _ Functions: Instructions that take zero or more arguments, do some computation, and return a value and/or have other sideeffects Examples: print(), len(), int(), raw_input()

8 Data Structures an organized form, such as an array list or string, in which connected data items are held in a computer some types: array, dictionary, graph, hash, heap, linked list, matrix, object, queue, ring, stack, tree, vector.

9 Variables a value that may change depending on the scope of a particular operation, problem, or set of operations ex. x in: x<-c(1:10) x<-c(10:1) x may be referred to independently of the function, can be a name, ex. agerange<-c(1:10)

10 Vectors Vectors are most basic data structure in R. Vectors are one-dimensional, sequentially ordered sequences of elements (numbers or character strings, or variables). Many other types of data structures can be understood in terms of vectors. ex. [1], [1,2,3], [a,b,c], sqrt(5), myfile.rtf square brackets tell us the position of each element in a vector:

11 Lists more complex than vectors may contain one or more tables, texts, numbers, or data structures best read into R from a separate file e.g., list of corpora you wish to search using an R function, strip demographic data from, replace one POS tag with another

12 Factors Similar to vectors, but... [NK] (p. 45) Gries mentions that "factors are internally represented as numbers". It appears that factors are similar to what I know from other languages as an enumeration, and are preferable to vectors due to memory optimization; if you are creating a data object with several data points but a limited set of values, it is more efficient to store the data points as integers along with a mapping from the integers to the set of values, rather than duplicating the values themselves for all identical data points. How do we think of them in Linguistics? From Statistics, a grouping variable. Each observation in a dataset, takes value representing defined levels of a categorical independent or dependent variable. (e.g., Gender - 2 levels - M/F) vectors may be factorized

13 Data types integers: 7, 2, -5 real numbers: , strings: hello, bom dia, 7 lists: [-7, 2, 5], [ hello, aloha ] in R, all elements of a vector must be of a single (the same) data type

14 Functions may take null arguments may take (upto) a particular number of arguments may require arguments to be labeled may have default settings for arguments, which will be used even when not explicitly stated by the user

15 file.choose() Do This: look this function up in R help What does it do? How many arguments does it take? What is the default setting of that argument? Useful because essentially removes need to set working directory for each R session Mac: file.choose() Windows: choose.files() Mac: file.list<-select.list(dir(scan(nmax=1, what= char )),multiple=t)

16 file.choose() #I can include comments with # #because R ignores everything after # my.variable <- c() str() #assignment occurs this way #examine the file structure length() nchar() getwd() quit() sqrt(6);sqrt(16) #get working directory #sometimes arg can be null #semicolons allow two functions to be #combined in one single line sample(100) #nice way to get a randomized list of #integers b/w 1 and 100 w/o replacement table(i) #yields a frequency table of a vector s elements

17 scan(file="", what=double(0),sep="") #loads the contents of text files into a vector what #double(0) if input is to be vector of numbers # char if input is to be vector of characters sep # any whitespace character can separate entries #\t indicates that tabs separate your entries #\n indicates that newlines (return) separates them scan(file="clipboard") #load the contents of the clipboard Don t forget to assign the output of scan to a variable! apply() #family of functions. This one works on arrays. A helpful link on the apply() functions is this:

18 Importing data Use read.table() to import data from an external file into a data frame Data frames are data structures They have row and column labels read.table(file.choose()) We still have to name it something to manipulate it in R ex. x<-read.table(file.choose())

19 Exporting data Use write.table() to export data to a file (at your user directory s root folder, unless specified otherwise) from a data frame you create in R Your file will have row and column labels ex. write.table(x,file="myrdataframe.txt")

20 More function help Sociolab wiki list of R functions: index.php?title=r Penn R Study Group: Mary Beckman s R scripts: ling600.01/scripts/