Rtutorial
Download as PPT, PDF2 likes754 views
This document provides an outline and overview of key topics in R programming including communication with R, R software, interfaces, code, packages, graphics, parallel processing, and commercial R products. It discusses how R has become a common language for statistics and data analysis. It provides examples of R code for basic math operations, vectors, sampling, and the central limit theorem. It also summarizes several R packages for bootstrapping, plotting, genomics, background correction, and parallel processing.
![R code
> 2+2
[1] 4
> 2+2^2
[1] 6
> (2+2)^2
[1] 16
> sqrt(2)
[1] 1.414214
> log(2)
[1] 0.6931472
>x=5
> y = 10
> z <- x+y
>z
[1] 15](https://image.slidesharecdn.com/rtutorial-140121072313-phpapp01/85/Rtutorial-7-320.jpg)
![R Code
> seq(1,5, by=.5)
[1] 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
> v1 = c(6,5,4,3,2,1)
> v1
[1] 6 5 4 3 2 1
> v2 = c(10,9,8,7,6,5)
>
> v3 = v1 + v2
> v3
[1] 16 14 12 10 8 6](https://image.slidesharecdn.com/rtutorial-140121072313-phpapp01/85/Rtutorial-8-320.jpg)
![R code
> max(v3);min(v3)
[1] 16
[1] 6
> length(v3)
[1] 6
> mean(v3)
[1] 11
> sd(v3)
[1] 3.741657](https://image.slidesharecdn.com/rtutorial-140121072313-phpapp01/85/Rtutorial-9-320.jpg)
![R code
> v4 = v3[v3>10]
> v4
[1] 16 14 12
> n = 1:10000; a = (1 + 1/n)^n
> cbind(n,a)[c(1:5,10^(1:4)),]
n
a
[1,] 1 2.000000
[2,] 2 2.250000
[3,] 3 2.370370
[4,] 4 2.441406
[5,] 5 2.488320
[6,] 10 2.593742
[7,] 100 2.704814
[8,] 1000 2.716924
[9,] 10000 2.718146](https://image.slidesharecdn.com/rtutorial-140121072313-phpapp01/85/Rtutorial-10-320.jpg)
![R code
# Birthday Problem
m = 100000; n = 25 # iterations; people in room
x = numeric(m)
# vector for numbers of matches
for (i in 1:m)
{
b = sample(1:365, n, repl=T) # n random birthdays in ith room
x[i] = n - length(unique(b)) # no. of matches in ith room
}
mean(x == 0); mean(x)
# approximates P{X=0}; E(X)
cutp = (0:(max(x)+1)) - .5
# break points for histogram
hist(x, breaks=cutp, prob=T) # relative freq. histogram](https://image.slidesharecdn.com/rtutorial-140121072313-phpapp01/85/Rtutorial-13-320.jpg)
Rtutorial
- 1. Tutorial on “R” Programming Language Eric A. Suess, Bruce E. Trumbo, and Carlo Cosenza CSU East Bay, Department of Statistics and Biostatistics
- 2. Outline • • • • • • • • Communication with R R software R Interfaces R code Packages Graphics Parallel processing/distributed computing Commerical R REvolutions
- 3. Communication with R • In my opinion, the R/S language has become the most common language for communication in the fields of Statistics and and Data Analysis. • Books are being written now with R presented directly placed within the text. • SV use R, for example • Excellent for teaching.
- 4. R Software • To download R • http://www.r-project.org/ • CRAN • Manuals • The R Journal • Books
- 5. R Software
- 6. R Interfaces • • • • • • • RWinEdt Tinn-R JGR (Java Gui for R) Emacs + ESS Rattle AKward Playwith (for graphics)
- 7. R code > 2+2 [1] 4 > 2+2^2 [1] 6 > (2+2)^2 [1] 16 > sqrt(2) [1] 1.414214 > log(2) [1] 0.6931472 >x=5 > y = 10 > z <- x+y >z [1] 15
- 8. R Code > seq(1,5, by=.5) [1] 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 > v1 = c(6,5,4,3,2,1) > v1 [1] 6 5 4 3 2 1 > v2 = c(10,9,8,7,6,5) > > v3 = v1 + v2 > v3 [1] 16 14 12 10 8 6
- 9. R code > max(v3);min(v3) [1] 16 [1] 6 > length(v3) [1] 6 > mean(v3) [1] 11 > sd(v3) [1] 3.741657
- 10. R code > v4 = v3[v3>10] > v4 [1] 16 14 12 > n = 1:10000; a = (1 + 1/n)^n > cbind(n,a)[c(1:5,10^(1:4)),] n a [1,] 1 2.000000 [2,] 2 2.250000 [3,] 3 2.370370 [4,] 4 2.441406 [5,] 5 2.488320 [6,] 10 2.593742 [7,] 100 2.704814 [8,] 1000 2.716924 [9,] 10000 2.718146
- 11. R code # LLN cummean = function(x){ n = length(x) y = numeric(n) z = c(1:n) y = cumsum(x) y = y/z return(y) } n = 10000 z = rnorm(n) x = seq(1,n,1) y = cummean(z) X11() plot(x,y,type= 'l',main= 'Convergence Plot')
- 12. R code # CLT n = 30 k = 1000 # sample size # number of samples mu = 5; sigma = 2; SEM = sigma/sqrt(n) x = matrix(rnorm(n*k,mu,sigma),n,k) # This gives a matrix with the samples # down the columns. x.mean = apply(x,2,mean) x.down = mu - 4*SEM; x.up = mu + 4*SEM; y.up = 1.5 hist(x.mean,prob= T,xlim= c(x.down,x.up),ylim= c(0,y.up),main= 'Sampling distribution of the sample mean, Normal case') par(new= T) x = seq(x.down,x.up,0.01) y = dnorm(x,mu,SEM) plot(x,y,type= 'l',xlim= c(x.down,x.up),ylim= c(0,y.up))
- 13. R code # Birthday Problem m = 100000; n = 25 # iterations; people in room x = numeric(m) # vector for numbers of matches for (i in 1:m) { b = sample(1:365, n, repl=T) # n random birthdays in ith room x[i] = n - length(unique(b)) # no. of matches in ith room } mean(x == 0); mean(x) # approximates P{X=0}; E(X) cutp = (0:(max(x)+1)) - .5 # break points for histogram hist(x, breaks=cutp, prob=T) # relative freq. histogram
- 14. R help • help.start() Take a look – An Introduction to R – R Data Import/Export – Packages • data() • ls()
- 15. R code Data Manipulation with R (Use R) Phil Spector
- 16. R Packages • There are many contributed packages that can be used to extend R. • These libraries are created and maintained by the authors.
- 17. R Package - simpleboot mu = 25; sigma = 5; n = 30 x = rnorm(n, mu, sigma) library(simpleboot) reps = 10000 X11() median.boot = one.boot(x, median, R = reps) #print(median.boot) boot.ci(median.boot) hist(median.boot,main="median")
- 18. R Package – ggplot2 • The fundamental building block of a plot is based on aesthetics and facets • Aesthetics are graphical attributes that effect how the data are displayed. Color, Size, Shape • Facets are subdivisions of graphical data. • The graph is realized by adding layers, geoms, and statistics.
- 19. R Package – ggplot2 library(ggplot2) oldFaithfulPlot = ggplot(faithful, aes(eruptions,waiting)) oldFaithfulPlot + layer(geom="point") oldFaithfulPlot + layer(geom="point") + layer(geom="smooth")
- 20. R Package – ggplot2 Ggplot2: Elegant Graphics for Data Analysis (Use R) Hadley Wickham
- 21. R Package - BioC • BioConductor is an open source and open development software project for the analysis and comprehension of genomic data. • http://www.bioconductor.org • Download > Software > Installation Instructions source("http://bioconductor.org/biocLite.R") biocLite()
- 22. R Package - affyPara library(affyPara) library(affydata) data(Dilution) Dilution cl <- makeCluster(2, type='SOCK') bgcorrect.methods() affyBatchBGC <- bgCorrectPara(Dilution, method="rma", verbose=TRUE)
- 23. R Package - snow • Parallel processing has become more common within R • snow, multicore, foreach, etc.
- 24. R Package - snow • Birthday Problem simulation in parallel cl <- makeCluster(4, type='SOCK') birthday <- function(n) { ntests <- 1000 pop <- 1:365 anydup <- function(i) any(duplicated( sample(pop, n,replace=TRUE))) sum(sapply(seq(ntests), anydup)) / ntests} x <- foreach(j=1:100) %dopar% birthday (j) stopCluster(cl) Ref: http://www.rinfinance.com/RinFinance2009/presentations/UIC-Lewis%204-25-09.pdf
- 25. REvolution Computing • REvolution R is an enhanced distribution of R • Optimized, validated and supported • http://www.revolution-computing.com/