R Programming for Data Science: Tutorial Guide for beginners

Introduction to R for Data Science

R programming for data science, a free programming language used by millions of people around the world for data analysis and statistics, is an important part of Data Science. R is a top choice for many data scientists and statisticians which has made it very popular. But what makes R so popular? Why and how should you use R for Data Science?"

R programming language, developed by Ross Ihaka and Robert Gentleman in 1993, is widely used for applications related to data science. R provides support for an extensive suite of statistical methods, inference techniques, machine learning algorithms, time series analysis, data analytics, and graphical plots to list a few. These features make it a great language for data exploration and investigation. 

Before we proceed further with programming in r for data science and what is r for data science? Let’s first discuss what is data science and what is a data scientist.

What is data science?

Data science is the study of data that involves developing methods of analyzing, recording and storing data to effectively extract useful information.The main aim of data science is to get in-depth knowledge about any type of structured and unstructured data.

What is a data scientist?

A data scientist is one who has technical skills to solve complex problems and who has curiosity to explore what kind of problems are needed to be solved. The main goal of data scientists is to analyze, process, and model data then interpret the outcomes to create actionable plans for companies and other organizations.

Data Science Training - Using R and Python

• Detailed Coverage
• Best-in-class Content
• Prepared by Industry leaders
• Latest Technology Covered

Download Curriculum

Getting started with R for Data Science

R can be downloaded from CRAN , the comprehensive R archive network. CRAN comprises a set of mirror servers distributed around the world and is used to distribute R and R packages. A new major version of R comes out once a year, and there are 2 to 3 minor versions each year. RStudio provides an integrated development environment, or IDE, for R programming.

What is Programming in R for Data Science?

Let us take a look at a simple program in R which prints “Hello World”. This can be accomplished either from the command line in the R interpreter or via a R script. Let us look at both mechanisms.

Hello World in R – from the R command prompt:

> print("Hello world!")

[1] "Hello world!"

Creating a HelloWorld.R script in R:

helloStr <- "Hello world!"

print(helloStr)

The script can be executed using Rscript HelloWorld.R. It will print:

[1] "Hello world!

What are different BuiltIn DataSets in R?

R comes with a large number of built in datasets.These can be used as demo data for understanding R packages and functions. Data sets in package 'datasets' include:

We can view the contents of any of these datasets using the following commands:

# Loading 
data() 
# Print the first n rows 
head(, )
For example, lets explore the iris datset:
data(iris)
head(iris,3)

What is Data ingestion in R?

R is used for data processing and analysis tasks. The first step in this activity is importing data. Base R provides several functions for this purpose. Let us explore some of these for a better understanding:

read.table : Used for importing tab delimited tabular data.e.g.,

  > df <- read.table("data.txt", header = FALSE)> df

read.csv : Used for importing csv file with comma(,) delimiter.e.g.,

read.csv2 : Used for importing csv file with semicolon(;) delimiter.

read.delim : Used for importing delimited file with any arbitrary delimiter.

Other library functions are available for importing data of specific format:e.g.,

• readxl:read_excel for reading excel files
• rjson:fromJSON for reading JSON data
• XML:xmlTreeParse for xml data.
• RCurl:readHTMLTable for reading HTML table data

What is Data Preparation and Cleansing in R?

Eighty percent of data analysis is spent on the cleaning and preparation of data. It is not just the first step, but may need to be repeated many times over the course of analysis.

In tidy data:

• Each variable forms a column.
• Each observation forms a row.
• Each type of observation unit forms a table.

Tidy represents a standard way of structuring a dataset. Real world datasets need not necessarily be available in tidy format:

• Column headers may not be variable names.
• Multiple variables might be stored in one column.
• Variables might be stored in rows.
• Unrelated observations might be stored in the same table.
• A single observational unit might be stored across multiple tables.

R provides a package tidyr for converting data into tidy format. tidyr provides three main functions for tidying up messy data:

• gather(),
• separate()and
• spread().

gather() takes multiple columns, and organizes them into key-value pairs.

For example, consider the dataset below which represents the test scores in 2 tests(a and b) for 3 individuals named Amar, Akbar and Anthony:

library(tidyr)

messy <- data.frame( name = c("Amar", "Akbar", "Anthony"), a = c(56, 91, 88), b = c(72, 64, 60) ) messy #>    name    a  b
#> 1  Amar   56 72
#> 2 Akbar   91 64
#> 3 Anthony 88 60

But ths dataset is currently not in a tidy format (Variables must correspond to columns). For it to be converted it into column format the data must be represented as name , test , score.

Let us see how we can use tidyr package to convert the existing dataset into tidy form.

messy %>%
  gather(test, score, a:b)
#>      name   test     score
#> 1   Amar    a        56
#> 2   Akbar   a        91
#> 3   Anthony a        88
#> 4   Amar    b        72
#> 5   Akbar   b        64
#> 6   Anthony b        60

Here we used the pipe operator %>%. The pipe operator allows you to pipe the output from one function to the input of another function. In our case the messy dataframe is piped as input to the gather function.

Similarly, separate function allows us to separate two variables are clumped together in one column.

spread(), takes two columns (key-value pair) and spreads them in to multiple columns, making data wider.

The tidied dataset can then be transformed as per the requirement of analysis.

R provides several packages for data transformation. Let us look at one of these – dplyr.

Below are some of the functions which are useful for this purpose:

• filter : Pick observations by their values
• arrange : Reordering the rows
• select : Pick variables by their names
• mutate : Create new variables in terms of functions of existing variables
• summarise : Create a single summary value from multiple given values
• group_by() : grouping operations in the “split-apply-combine” concept

For example let us determine all the entries in the iris datset with Species as ‘virginica’ and Sepal.Width:

> library(dplyr)

> filter(iris,Species=="virginica",Sepal.Width>3)

Sepal.Length Sepal.Width Petal.Length Petal.Width Species

What is Data Modeling in R?

A model provides a simple low-dimensional summary of a given dataset. R provides inbuilt functions that make fitting statistical models very simple.

The function to fit linear models is called lm. It is very useful for regression analysis of dataset.The generic syntax is as follows:

> fitted_model <- lm(formula, data = data.frame) For example: > fitted_model <- lm(y ~ x1 + x2 + x3, data = production) Will fit a multiple regression model of dependent variable y on independent variables x1,x2 and x3. Generalized linear models extend linear models to distributions such as gaussian, binomial, poisson, inverse gaussian,gamma and quasi-likelihood models. > fitted_model <- glm(formula, family=family.generator, data=data.frame) `For example: > fitted_model <- glm(y ~ x1 + x2, family = gaussian, data = production) Or > fitted_model <- glm(y ~ x, family = binomial(link=probit), data = mydata) Or > fitted_model <- glm(y ~ x1 + x2 - 1, family = quasi(link=inverse, variance=constant), data = testdata) The model parameters can be visualized by calling > summary(fitted_model)

Example:
## 
## Call:
## glm(formula = formula, family = "binomial", data = mydata)
## ## Deviance Residuals: 
##     Min       1Q   Median       3Q      Max  
## -2.6456  -0.5858  -0.2609  -0.0651   3.1982  
## 
## Coefficients:
##                           Estimate Std. Error z value Pr(>|z|)    
## (Intercept)                0.07882    0.21726   0.363  0.71675    
## age                        0.41119    0.01857  22.146  

Besides these R also provides support for other models such as :

• Classification and Regression model – caret package
• Mixed models – nlme package
• Robust Regression – package MASS ( removes outliers)
• Additive models – package acepack
• Tree models – package rpart,tree

What is Data Visualization in R?

Data visualization is an important aid in data analysis and decision making.ggplot2 is a data visualization package for R. ggplot2 is an implementation of Grammar of Graphics(gg)—a general scheme for data visualization which breaks up graphs into components such as scales and layers. In contrast to base R graphics using plot function, ggplot2 allows the user to add, remove or alter components in a plot at a high level of abstraction.

ggplot(dat, aes(year, lifeExp)) + geom_point()

This will create a graph between year and life expectancy data from the dataset dat and depict it using geometric points on the graph.

Different types of plots can be created by making use of additional graphing primitives such as geom_lines(),geom_boxplot(),geom_smooth() etc.

qplot is a convenient wrapper on tip of ggplot2 for creating a number of different types of plots .

The generic syntax for qplot is :

qplot(x, y, ..., data, facets = NULL, margins = FALSE, geom = "auto", 
xlim = c(NA, NA), ylim = c(NA, NA), log = "", main = NULL, xlab = NULL,
ylab = NULL, asp = NA, stat = NULL,position = NULL)

e.g.: qplot(mpg, wt, data = mtcars)

f <- function() {
   a <- 1:10
   b <- a ^ 3
   qplot(a, b)
}
f()

This will plot a curve with a[1-10] on x-axis and b=a^3 on y axis and the (x,y) pairs being represented by points.  

“You can be an R-programming professional by Enrolling Today”

Data Science Training - Using R and Python

• Personalized Free Consultation
• Access to Our Learning Management System
• Access to Our Course Curriculum
• Be a Part of Our Free Demo Class

Sign Up Now

Most Common R Libraries?for Data Science

R has many packages and libraries to help with different tasks in Data Science. Here are some of the best add-on packages for R recommended by RStudio:

1. Database Packages

• DBI: Connects R with different databases.
• RMySQL and RSQLite: Load and read data from databases like MySQL and SQLite.

2. Visualization Packages

• ggplot2: Creates beautiful and easy-to-understand plots and graphs.
• ggmap: Downloads maps from Google Maps and adds them to your plots.
• shiny: Helps you build web apps.

3. Data Manipulation and Analysis Packages

• dplyr: Makes it easy to summarize and rearrange data.
• stringr: Provides simple tools to work with text.
• lubridate: Helps you work with dates and times in your data.
• DataExplorer: Useful for exploring data, creating new features, and making reports.

4. Machine Learning and Deep Learning Packages

• randomForest and caret: Train models to predict outcomes.
• deepnet: Tools for deep learning. You can also use popular frameworks like Keras and TensorFlow with R.
• devtools: Helps you create your own R packages.

Conclusion:

R as a language is developed from ground up for data analysis and data interpretation. As is rightly said, data represents power in the new economy. But we need appropriate tools to harness the power inherent in raw data. R programming for data science provides us with this power. With an ever growing user community and expanding package list covering all facets of data science, R is a language of choice for data science. This post provides a brief introduction to R and its capabilities so that readers can get started quickly and begin exploring further all the powerful features available for data modelling and interpretation.