Hands-on course on July 1 2015 07-01-2015d1710

2015-07-01

Hands-on course on July 1 2015 07-01-2015d1710



some notes 


07-01-2015d1012





R Statistics course





Ramon Diaz-Uriarte





do not read Gentleman's bioinfo one, nor Crawley's





RCommander


JGR (Java Gui for R)


JGR + Deducer





runif = random uniform distribution





R variables can hold any type of object (this does not need to be explicitly defined)





rnorm


create a random number from a normal distribution from 1 or 0 to a max value





rep is for repeat


c is for concatenate (paste together)


factor object is good when you have something that is a class label in a statistical analysis.





head function shows first 6 lines of an object





apply function applies something to an object


apply(object, dimension that you will do things, )





1:20 defines a range from 1 to 20





functions can be defined on the fly while being passed as a parameter into a function





thin line between using and programming R





names tells you the names of each component in a list





The $ sign used to access the interal components of a list.





functional composition





library to load a library/package in library





It's better not to save the workspace from RStudio





? in front of function brings up help





you can type example around a function to see an example





run





he likes to be explicit about using the names of variables being passed to a function





apropos searches functions that have certain text in their name.





vignettes give you several work examples (pdf user guides)





sos package for using help





cran task views





coding style





try not to go beyond column 80





variables are really a vector in r





vectors hold elements of the same type.  can't mix numbers and strings





if a vector has more than 1 dimension, then it is a matrix





dataframes can have elements of different types


dataframes organize things in terms of columns


like a spreadsheet


dataframes are rectangular table objects





dataframes are a type of list





a list is a general container





getwd() is get working directory





the working directory can also be changed (session->set working directory)





setwd()








summary of a data frame





str of a data frame (str is for structure)





read.table function  (the header parameter is VERY important to indicate and pay attention to)





dealing with missing values; use an NA





read csv function to get csv file





You can save specific objects in R





ggplot2





source command to run a script





vectors


regular sequences with sequence function





range by 1 like 2:7





rep function (repeat function)





output often length of largest object





recycling rule can lead to nasty surprises





he doesn't like using "=" for assignment





identical function





floating point arithmetic in r (be careful)





which can give indices of elements in a list that meet criteria





to get elements of a vector pass another vector with the positions you want





You can name each element in a vector in R


ages <- c(Juan = 23, Maria = 35, Irene = 12, Ana = 93)


vectors that have names are not dataframes


this is like a lookup table





age





factors


factors are recoded to numeric things


as.numeric()





rbind to bind rows


cbind to bind columns





you can use rbind to add something to a matrix





drop = FALSE to not drop dimensions





the apply function can be used on a matrix, but not on a vector





obtaining indices of an object of a matrix





which(A==999, arr.ind = TRUE)


     row col


[1,]   2   3








a list is a very general container





s3 and s4 classes allow objects to have functions to allow for a type of object oriented programming


s4 classes are more sophisticated whereas s3 is for things like "print", "plot", (more built-in things), etc.





A dataframe is a special type of list





transform datafram into matrix


data.matrix(AB)


as.matrix(AB)





with function to get data from tables





attach can add things to the list of packages that R searches through when finding a command, but this strategy is not used often anymore.





iteration in R


names.of.friends <- c("Ana", "Rebeca", "Marta", 


                      "Quique", "Virgilio")


for(friend in names.of.friends) {


  cat("\n I should call", friend, "\n")


apply can be used instead of for loops and they are easier to parallelize

defining a function in R example 


multByTwo <- function(x) {


  z <- 2 * x


  return(z)

}

optional arguments in R example 


plotAndLm <- function(x, y, title = "A figure") {


  lm1 <- lm(y ~ x)


  cat("\n Printing the summary of x\n")


  print(summary(x))


  cat("\n Printing the summary of y\n")


  print(summary(y))


  cat("\n Printing the summary of the linear regression\n")


  print(summary(lm1))


  plot(y ~ x, main = title)


  abline(lm1)


  return(lm1)


}
}