A Taste of R

A Taste of R

R is widely accepted as a powerful as well as easy-to-learn language in data analysis, though python has got its own numpy, pandas, sympy, matplotlib, scipy, and more handy modules, R can still serve as a sharp weapon to deal with data.

This passage is mainly a comprehension on P.Haschke’s R-Course. Some documents like official ones are included as well.

Basics

extremely basic operations shall not be included in this text. All comments start with #.

• To get help, type ?<anything>or help(<anything>) , ?log() equals but can’t replace help(log).

• To get functions can be used on anything, use aporopos("<anything>"). For example, apropos("mean"). Looks familiar, python got dir().

• By install.packages("<packageName>") better be conducted in the console, you can get libraries you want from the CRAN mirror, to use anything inside your module or session, import the library using library("<packageName>").

Introduction

• Objects types are: Vectors(containing elements of the same type, one-dimensional), Matrices & Arrays(two or more dimensional, same type), Lists(like vectors but different type allowed), Data Frames(like table mapping, two-dimensional), Factors, Functions(As you know)

• Modes types are: integer, numeric(real numbers), complex, character(AKA strings), logical(AKA Bool)

• Assignment: <variableName> <- <toBeAssigned>, to get the type, use is(variableName), to get and remove the variables, use ls() and rm(<variableToBeRemoved>). I have to admit this part is weird, bash and lambda?

• function c() is for concatenating elements making vectors.

• some new(to me) functions may be sd()(standard deviation, var()(variance), cov()(covariance), cor()(correlation coefficient), unique() and which is rather tricky. prod() is awesome. The most weird thing I believe is the use and access of variables like <var>.<v> and <var>$<v>.

• seq() is used for creating defined vectors, rep() do repeat. Notation usage is similar to python but more powerful, simply Vector[Vector >= 1] can save you lots of trouble. summary() shows basic information you need. subset() works perfect like map() function in python.

• print(), paste(), cat() will deal with Stdout, paste() is for multiple modes, cat() doesn’t create object in active memory

Matrices

• source() import your previous codes. save() saves your changes.

• Matrices have three main arguments: data(R object), nrow(number of rows), ncol(number of columns). Two more optional arguments are byrow and dimnames.

• rbind() & cbind() can be useful. rownames() & colnames() cane be used to define and get names. diag(), you knew it already.

• Matrices use [] for notations, black magic like Matrix[Matrix[ , 2] > 4, ] can be astonishing. t() for transpose, solve() for inverse, det() for determinant, chol() for cholesky decomposition, eigen() is perfect, crossprod() for cross product. %*% for matrix multiplication.

Data Frames

• data() shows and load packages; First library(), then data(). class() like is(), names() is self-explanatory.

• read.csv() for csv, download.file() for getting files. DataFram$name is for particular name extraction.

• str() and describe() for summary.

Graphics

• ggplot2 shall be installed in advance, and to set more, do like this plot1 <- plot1 + geom_point(aes(x = FEhighway, y = FEcity)), pretty ugly to me. :( .Other options can be referred at ggplot

Programs

• if & ifelse() , other control flow functions just like C-style in one-line

• Defined functions are javascript-style, consider this

  	```r
  	MyFunction <- function(Object) {
  		Object + Object
  	}
  	```