Introduction to R

1
Introduction to R

• A. Di Bucchianico

2
Types of statistical software

• command-line software
• requires knowledge of syntax of commands
• reproducible results through scripts
• detailed analyses possible
• GUI-based software
• does not require knowledge of commands
• not reproducible actions
• hybrid types (both command-line and GUI)

3
Well-known statistical software

• SAS
• SPSS
• Minitab
• Statgraphics
• S-Plus
• R

4
R

• free
• language almost the same as S
• maintained by top quality experts
• available on all platforms
• continuous improvement
• Available through www.r-project.org

5
Contents

• Basic operations
• Data creation I/O
• Component extraction
• Plots
• Basic statistics
• Libraries
• Regression analysis
• Survival analysis

6
Basic operations

• assignment operation a lt- 2sqrt(5)
• help function
• help(pnorm)
• help.search(normal distribution)
• probability functions
• d (density) dgamma(x,n,?)
• p (probabilitycdf) pweibull(x,3,2)
• q (quantile) qnorm(0.95)
• r (random numbers) rexp(10,?)

7
Data creation I/O

• create
• vectors c(1,2,3)
• matrices matrix(c(1,2,3,4,5,6),2,3,byrowT)
  (2rows)
• list
• patterns
• (1,2,3) 13
• seq (1,2,3) seq(1,3,by1)
• working directories and files
• setwd
• getwd
• attach
• read data
• from file read.table(file.txt,headerTRUE)
• from web read.data.url

8
Component extraction

• dr, rth row of object d
• d,c cth column of object d
• dr,c entry in row r and column c of object d
• length(d) length of d
• ddlt20 extract all elements of d that are
  smaller than 20
• dage extract column age from object d

9
Plots

• plot both 1D and 2D plots
• hist histogram
• qqnorm normal probability plot
  (quantile-quantile plot)
• Save graphics by choosing File -gt Save as

10
Basic statistics

• summary
• mean
• stdev
• t.test
• boxplot

11
Packages

• specialized functions available through packages
  and libraries
• in Windows interface choose Packages -gt Load
  Packages
• examples of packages
• qcc (quality control)
• survival

12
Functions

• Analyses that have to be performed often can be
  put in the form of functions
• Example simple lt- function(data,mean0,alpha0.05
  )
• hist(data),t.test(data,conf.levelalpha,mumean,a
  lternativetwo-sided)
• simple(data,4) uses the default value 0.05 and
  test the null hypothesis mu4.

13
Regression analysis

• general command lm (linear model)
• requires data to be available in the form of a
  data frame
• more general than matrix because columns need not
  have same length)
• use command data.frame for conversion
• other types of regression also possible (see also
  dedicated packages)

14
Survival analysis

• through library Surv of survival
• Cox proportional hazards coxph

15
Useful web sites

• www.r-project.org
• http//cran.r-project.org/doc/contrib/Short-refcar
  d.pdf
• http//www.uni-muenster.de/ZIV/Mitarbeiter/BennoSu
  eselbeck/s-html/shelp.html
• http//www.maths.lth.se/help/R/
• http//www.mas.ncl.ac.uk/ndjw1/teaching/sim/R-int
  ro.html
• http//stats.math.uni-augsburg.de/JGR/
• http//socserv.mcmaster.ca/jfox/Misc/Rcmdr/index.h
  tml