R Introduction.md

R Introduction

author: Sixiang Hu date: 30th Nov, 2014 width: 1280 height: 1024 autosize: true

What is R?

• The R statistical programming language is a free open source package based on - the S language developed by Bell Labs.
• Many statistical functions are already built in. Contributed packages expand the functionality to cutting edge research.
• Free

Getting Started

• Download R 3.1.2 (Pumpkin Helmet):

  • http://cran.r-project.org/

• Download RStudio (Optional):

  • http://www.rstudio.com/products/rstudio/download/
  • To get a beautiful user interface
  • Easier maintains

Basic Assignment and Operations

• Arithmetic Operations:

1+2*exp(3)/sin(4)

[1] -52.08

• Assignment:

a <- runif(4)
a

[1] 0.1926935 0.1130957 0.7909896 0.3856126

• Use help in R:

?sin

Variables in R

There are several types of variables can be used in R: integer, numerical, character, date, vector, matrix, list, and data frame.

a <- sample(LETTERS,5,replace=T)
a

[1] "P" "C" "E" "Y" "L"

class(a)

[1] "character"

class(data.frame(c(a=a,b=a)))

[1] "data.frame"

Data Frame

Data frame is the widest used data type used in R. It can have different type of value in different columns, but values in a column must keep the same.

a <- c(1,2,3)
b <- c("a","b","c")
df <- data.frame(a=a,b=b)
df

  a b
1 1 a
2 2 b
3 3 c

df$a

[1] 1 2 3

df[2,2]

[1] b
Levels: a b c

List

List can be used to store different types of variables with differnet length.

a <- c(1,2,3)
b <- c("a","b","c")
lt <- list(a,b)
lt

[[1]]
[1] 1 2 3

[[2]]
[1] "a" "b" "c"

lt[[2]]

[1] "a" "b" "c"

Function

We can create functions that can be used repetitely:

my_square <- function(x) {x*x}
my_square(5)

[1] 25

Plot

par(mfrow=c(2,2))
plot(cars,type="l")
hist(cars$speed)
plot(cars)
hist(cars$dist)

Modelling

 model1 <- glm(mpg~gear+cyl,data=mtcars,family=Gamma)
summary(model1)

Call:
glm(formula = mpg ~ gear + cyl, family = Gamma, data = mtcars)

Deviance Residuals: 
     Min        1Q    Median        3Q       Max  
-0.34679  -0.08718  -0.01579   0.08210   0.25549  

Coefficients:
              Estimate Std. Error t value Pr(>|t|)    
(Intercept)  0.0186841  0.0115534   1.617    0.117    
gear        -0.0020063  0.0021761  -0.922    0.364    
cyl          0.0067352  0.0008912   7.558 2.48e-08 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

(Dispersion parameter for Gamma family taken to be 0.02356935)

    Null deviance: 2.73529  on 31  degrees of freedom
Residual deviance: 0.70551  on 29  degrees of freedom
AIC: 166.14

Number of Fisher Scoring iterations: 4

We can use different family and link function for GLM modeling. Interactions, offset, splines and transformation are all can be done within R.

For some large dataset (>100MB), it will be slow for a fit. Some packages bigGLM can be used to accelarate to process.

Advanced Plot (ggplot2)

library(ggplot2)
library(ggthemes)
ggplot(data = msleep, aes(x = log(bodywt), y = sleep_total)) +
  geom_point(aes(shape=vore), size=3)+
  theme_wsj()

Further Online Tutorial

• R Tutorial (Statistics)

  • http://www.r-tutor.com/

• CRAN R Tutorial

  • http://cran.r-project.org/doc/manuals/r-release/R-intro.html

• Data Camp R

  • https://www.datacamp.com/

• Kaggle

  • https://www.kaggle.com/