Just like doctors cannot treat all medical symptoms with generic paracetamol and ibuprofen, data scientists need to use more potent methods to store and manage vast amounts of data. Knowing already how Big Data can be defined, and what requirements have to be met in order to qualify data as Big, we can now take a step forward and introduce a number of tools that are specialized in dealing with these enormous data sets. Although traditional techniques may still be valid in certain circumstances, Big Data comes with its own ecosystem of scalable frameworks and applications that  facilitate the processing and management of unusually large or fast data. In this chapter, we will briefly present several most common Big Data tools, which will be further explored in greater detail later on in the book.

Hadoop - the elephant in the room

If you have been in the Big Data industry for as little as one day, you surely must have heard the unfamiliar sounding word Hadoop, at least every third sentence during frequent tea break discussions with your work colleagues or fellow students. Named after Doug Cutting's child's favorite toy, a yellow stuffed elephant, Hadoop has been with us for nearly 11 years. Its origins began around the year 2002 when Doug Cutting was commissioned to lead the Apache Nutch project-a scalable open source search engine. Several months into the project, Cutting and his colleague Mike Cafarella (then a graduate student at University of Washington) ran into serious problems with the scaling up and robustness of their Nutch framework owing to growing storage and processing needs. The solution came from none other than Google, and more precisely from a paper titled The Google File System authored by Ghemawat, Gobioff, and Leung, and published in the proceedings of the 19th ACM Symposium on Operating Systems Principles. The article revisited the original idea of Big Files invented by Larry Page and Sergey Brin, and proposed a revolutionary new method of storing large files partitioned into fixed-size 64 MB chunks across many nodes of the cluster built from cheap commodity hardware. In order to prevent failures and improve efficiency of this setup, the file system creates copies of chunks of data, and distributs them across a number of nodes, which were in turn mapped and managed by a master server. Several months later, Google surprised Cutting and Cafarella with another groundbreaking research article known as MapReduce: Simplified Data Processing on Large Clusters, written by Dean and Ghemawat, and published in the Proceedings of the 6th Conference on Symposium on Operating Systems Design and Implementation.

The MapReduce framework became a kind of mortar between bricks, in the form of data distributed across numerous nodes in the file system, and the outputs of data transformations and processing tasks.

The MapReduce model contains three essential stages. The first phase is the Mapping procedure, which includes indexing and sorting data into the desired structure based on the specified key-value pairs of the mapper (that is, a script doing the mapping). The Shuffle stage is responsible for the redistribution of the mapper's outputs across nodes, depending on the key; that is, the outputs for one specific key are stored on the same node. The Reduce stage results in producing a kind of summary output of the previously mapped and shuffled data, for example, a descriptive statistic such as the arithmetic mean for a continuous measurement by each key (for example a categorical variable). A simplified data processing workflow, using the MapReduce framework in Google and Distributed File System, is presented in the following figure:

A diagram depicting a simplified Distributed File System architecture and stages of the MapReduce framework

The ideas of the Google File System model, and the MapReduce paradigm, resonated very well with Cutting and Cafarella's plans, and they introduced both frameworks into their own research on Nutch. For the first time their web crawler algorithm could be run in parallel on several commodity machines with minimal supervision from a human engineer.

In 2006, Cutting moved to Yahoo! and in 2008, Hadoop became a separate Nutch independent Apache project. Since then, it's been on a never-ending journey towards greater reliability and scalability to allow bigger and faster workloads of data to be effectively crunched by gradually increasing node numbers. In the meantime, Hadoop has also become available as an add-on service on leading cloud computing platforms such as Microsoft Azure, Amazon Elastic Cloud Computing (EC2), or Google Cloud Platform. This new, unrestricted, and flexible way of accessing shared and affordable commodity hardware, enabled numerous companies as well as individual data scientists and developers to dramatically cut their production costs and process larger than ever data in a more efficient and robust manner. A few Hadoop record-breaking milestones are worth mentioning at this point. In the well-known real-world example at the end of 2007, the New York Times was able to convert more than 4TB of images, within less than 24 hours, using a cluster built of 100 nodes on Amazon EC2, for as little as $200. A job that would have potentially taken weeks of hard labor, and a considerable amount of working man-hours, could now be achieved at a fraction of the original cost, in a significantly shorter scope of time. A year later 1TB of data was already sorted in 209 seconds and in 2009, Yahoo! set a new record by sorting the same amount of data in just 62 seconds. In 2013, Hadoop reached its best Gray Sort Benchmark ( http://sortbenchmark.org/ ) score so far. Using 2,100 nodes, Thomas Graves from Yahoo! was able to sort 102.5TB in 4,328 seconds, so roughly 1.42TB per minute.

In recent years, the Hadoop and MapReduce frameworks have been extensively used by the largest technology businesses such as Facebook, Google, Yahoo!, major financial and insurance companies, research institutes, and Academia, as well as Big Data individual enthusiasts. A number of enterprises offering commercial distributions of Hadoop such as Cloudera (headed by Tom Reilly with Doug Cutting as Chief Architect) or Hortonworks (currently led by Rob Bearden-a former senior officer at Oracle and CEO of numerous successful open source projects such as SpringSource and JBoss, and previously run by Eric Baldeschwieler-a former Yahoo! engineer working with Cutting) have spun off from the original Hadoop project, and evolved into separate entities, providing additional Big Data proprietary tools, and extending the applications and usability of Hadoop ecosystem. Although MapReduce and Hadoop revolutionized the way we process vast quantities of data, and hugely propagated Big Data analytics throughout the business world and individual users alike, they also received some criticism for their still present performance limitations, reliability issues, and certain programming difficulties. We will explore these limitations and explain other Hadoop and MapReduce features in much greater detail using practical examples in Chapter 4, Hadoop and MapReduce Framework for R.

By now you have been introduced to the notion of Big Data, its features, and its characteristics, as well as the most widely used Big Data analytics tools and frameworks such as Hadoop, HDFS, MapReduce framework, relational and non-relational databases, and Apache Spark project. They will be explored more thoroughly in the next few chapters of this book, but now the time has finally come to present the true hero and the main subject of this book-the R language. Although R as a separate language on its own has been with us since the mid 90s, it is derived from a more archaic S language developed by John Chambers in the mid 1970s. During Chambers' days in Bell Laboratories, one of the goals of his team was to design a user-friendly, interactive, and quick-in-deployment interface for statistical analysis and data visualizations. As they frequently had to deal with non-standard data analyses, and different data formats, the flexibility of this new tool, and its ability to make the most of the previously used Fortran algorithms, were the highest-order priorities. Also, the project was developed to implement certain graphical capabilities in order to visualize the outputs of numeric computations. The first version was run on a Honeywell operated machine, which wasn't the ideal platform, owing to a number of limitations and impracticalities.

The continuous work on S language progressed quite quickly and by 1981 Chambers and his colleagues were able to release a Unix implementation of the S environment along with a first book-cum-manual titled S: A Language and System for Data Analysis. In fact S was a hybrid-an environment or an interface allowing access to Fortran-based statistical algorithms through a set of built-in functions, operating on data with the flexibility of custom programmable syntax, for those users who wished to go beyond default statistical methods and implement more sophisticated computations. This hybrid, quasi object-oriented and functional programming language-like, statistical computing tool created a certain amount of ambiguity and confusion, even amongst its original developers. It is a well-known story that John Chambers, Richard Becker, and other Bell Labs engineers working on S at that time had considerable problems in categorizing their software as either a programming language, a system, or an environment. More importantly, however, S found a fertile ground in Academia and statistical research, which resulted in a sharp increase in external contributions from a growing community of S users. Future versions of S would enhance the functional and object-based structure of the S environment by allowing users to store their data, subsets, outputs of the computations, and even functions and graphs as separate objects. Also, since the third release of S in 1986, the core of the environment has been written in the C language, and the interface with Fortran modules has become accessible dynamically through the evaluator directly from S functions, rather than through a preprocessor and compilation, which was the case in the earlier versions. Of course, the new releases of the modernized and polished S environment were followed by a number of books authored by Chambers, Becker, and now also Allan Wilks, in which they explained the structure of the S syntax and frequently used statistical functions. The S environment laid the foundations for the R programming language, which has been further redesigned in 1990s by Robert Gentleman and Ross Ihaka from the University of Auckland in New Zealand. Despite some differences (a comprehensive list of differences between the S and R languages is available at https://cran.r-project.org/doc/FAQ/R-FAQ.html#R-and-S ), the code written in the S language runs almost unaltered in the R environment. In fact R and also the S-Plus language are the implementations of S-in its more evolved versions, and it is advisable to treat them as such. Probably the most striking difference between S and R, comes from the fact that R blended in an evaluation model based on the lexical scoping adopted from Steele and Sussman's Scheme programming language. The implementation of lexical scoping in R allowed you to assign free variables, depending on the environment in which a function referencing such variables was created, whereas in S free variables could have only been defined globally. The evaluation model used in R meant that the functions, in comparison with S, were more expressive and written in a more natural way giving the developers or analysts more flexibility and greater capabilities by defining variables specific to particular functions within their own environments. The link provided earlier, lists other minor, or very subtle, differences between both languages-many of which are sometimes not very apparent unless they are explicitly referred to in error messages when attempting to run a line of R code.

Currently, R comes in various shapes or forms, as there are several open source and commercial implementations of the language. The most popular, and the ones used throughout this book, are the free R GUI available to download from the R Project CRAN website at https://cran.r-project.org/ and our preferred and also free-of-charge RStudio IDE available from https://www.rstudio.com/ . Owing to their popularity and functionalities, both implementations deserve a few words of introduction.

The most generic R implementation is a free, open source project managed and distributed by the R Foundation for Statistical Computing headquartered in Vienna (Austria) at the Institute for Statistics and Mathematics. It is a not-for-profit organization, founded and set up by the members of the R Development Core Team, which includes a number of well-known academics in statistical research, R experts, and some of the most prolific R contributors over the past several years. Amongst its members, there are still the original fathers of the S and R languages such as John Chambers, Robert Gentleman, and Ross Ihaka, as well as the most influential R developers: Dirk Eddelbuettel, Peter Dalgaard, Bettina Grun, and Hadley Wickham to mention just a few. The team is responsible for the supervision and the management of the R core source code, approval of changes and alterations to the source code, and the implementation of community contributions. Through the web pages of the Comprehensive R Archive Network (CRAN) ( https://cran.r-project.org/ ), the R Development Core Team releases new versions of the R base installation and publishes recent third-party packages created by independent R developers and R community members. They also release research-oriented, open access, refereed periodic R Journals, and organize extremely popular annual useR! conferences, which regularly gather hundreds of passionate R users from around the world.

The R core is the basis for enterprise-ready open source and commercial license products developed by RStudio operating from Boston, MA and led by the Chief Scientist Dr. Hadley Wickham-a creator of numerous crucial data analysis and visualization packages for R, for example, ggplot2, rggobi, plyr, reshape, and many more. Their IDE is probably the best and most user-friendly R interface currently available to R users, and if you still don't have it, we recommend that you install it on your personal computer as soon as possible. The RStudio IDE consists of an easy-to-navigate workspace view, with a console window, and an editor, equipped with code highlighting and a direct code execution facility, as well as additional views allowing user-friendly control over plots and visualizations, file management within and outside the working directory, core and third-party packages, history of functions and expressions, management of R objects, code debugging functionalities, and also a direct access to help and support files.

As the R core is a multi-platform tool, RStudio is also available to users of the Windows, Mac, or Linux operating systems. We will be using RStudio desktop edition as well as the open source version of the RStudio Server throughout this book extensively, so please make sure you download and install the desktop free version on your machine in order to be able to follow some of the examples included in Chapter 2, Introduction to R Programming Language and Statistical Environment and Chapter 3, Unleashing the Power of R from Within. When we get to cloud computing (Online Chapter, Pushing R Further, https://www.packtpub.com/sites/default/files/downloads/5396_6457OS_PushingRFurther.pdf) we will explain how to install and use RStudio Server on a Linux run server.

The following are screenshots of graphical user interfaces (on Mac OS X) of both the R base installation, available from CRAN, and the RStudio IDE for desktop. Please note that for Windows users, GUIs may look a little different than the attached examples; however, the functionalities remain the same in most cases. RStudio Server has the same GUI as the desktop version; there are however some very minor differences in available options and settings:

R core GUI for Mac OS X with the console panel (on the left) and the code editor (on the right)

RStudio IDE for Mac OS X with execution console, code editor panels, and other default views. The Cobalt theme was used in the preceding example

After many years of R being used almost exclusively in Academia and research, recent years have witnessed an increased interest in the R language coming from business customers and the financial world. Companies such as Google and Microsoft have turned to R for its high flexibility and simplicity of use. In July 2015, Microsoft completed the acquisition of Revolution Analytics-a Big Data and predictive analytics company that gained its reputation for their own implementations of R with built-in support for Big Data processing and analysis.

The growth of R does not surprise. Thanks to its vibrant and talented community of enthusiastic users, it is currently the world's most widely-used statistical programming language with nearly 8,400 third-party packages available in the CRAN (as of June 2016) and many more libraries featured on BioConductor, Forge.net, and other repositories, not to mention hundreds of packages under development on GitHub, as well as many other unachieved and accessible only through the personal blogs and websites of individual researchers and organizations.

Apart from the obvious selling points of the R language, such as its open source license, a lack of any setup fees, unlimited access to a comprehensive set of ready-made statistical functions, and a highly active community of users, R is also widely praised for its data visualization capabilities and ability to work with data of many different formats. Popular and very influential newspapers and magazines such as the Washington Post, Guardian, the Economist, or the New York Times, use R on a daily basis to produce highly informative diagrams and info graphics, in order to explain complex political events or social and economical phenomena. The availability of static graphics through extremely powerful packages such as ggplot2 or lattice, has lately been extended even further to offer interactive visualizations using the shiny or ggobi frameworks, and a number of external packages (for example rCharts created and maintained by Ramnath Vaidyanathan) that support JavaScript libraries for spatial analysis, for example leaflet.js, or interactive data-driven documents through morris.js, D3.js, and others. Moreover, R users can now benefit from Google Visualization API charts, such as the famous motion graph, directly through googleVis package developed by Markus Gesmann, Diego de Castillo, and Joe Cheng (the following screenshot shows that googleVis package in action):

An example of a motion chart created by the Google Visualization package-an R interface to Google charts

As mentioned earlier, R works with data coming from a large array of different sources and formats. This is not just limited to physical file formats such as traditional comma-separated or tab-delimited formats, but it also includes less common files such as JSON (a widely used format in web applications and modern NoSQL databases and that we will explore in detail in Chapter 6, R with Non-Relational and (NoSQL) Databases) or images, other statistical packages and proprietary formats such as Excel workbooks, Stata, SAS, SPSS, and Minitab files, scrapping data from the Internet, or direct access to SQL or NoSQL databases as well as other Big Data containers (such as Amazon S3), or files stored in the HDFS. We will explore most of the data import capabilities of R in a number of real-world examples throughout this book. However if you would like to get a feel for a variety of data import methods in R right now, please visit the CRAN page at https://cran.r-project.org/manuals.html, which lists a set of the most essential manuals including the R Data Import/Export document outlining the import and export methods of data to and from R.

Finally, the code in the R language can easily be called from other programming platforms such as Python or Julia. Moreover, R itself is able to implement functions and statements from other programming languages such as the C family, Java, SQL, Python, and many others. This allows for greater veracity and helps to integrate the R language into existing data science workflows. We will explore many examples of such implementations throughout this book.

In the first chapter we explained the ambiguity of Big Data definitions and highlighted its major features. We also talked about a deluge of Big Data sources, and mentioned that even one event, such as Messi's goal, can lead to an avalanche of large amounts of data being created almost instantaneously.

You were then introduced to some most commonly used Big Data tools we will be working with later, such as Hadoop, its Distributed File System and the parallel MapReduce framework, traditional SQL and NoSQL databases, and the Apache Spark project, which allows faster (and in many cases easier) data processing than in Hadoop.

We ended the chapter by presenting the origins of the R programming language, its gradual evolution into the most widely-used statistical computing environment, and the current position of R amongst a spectrum of Big Data analytics tools.

In the next chapter you will finally have a chance to get your hands dirty and learn, or revise, a number of frequently used functions in R for data management, transformations, and analysis.