Laura Igual - Introduction to Data Science

Departament de Matemtiques i Informtica, Universitat de Barcelona, Barcelona, Spain

Laura Igual

Email:

You have, no doubt, already experienced data science in several forms. When you are looking for information on the web by using a search engine or asking your mobile phone for directions, you are interacting with data science products. Data science has been behind resolving some of our most common daily tasks for several years.

Most of the scientific methods that power data science are not new and they have been out there, waiting for applications to be developed, for a long time. Statistics is an old science that stands on the shoulders of eighteenth-century giants such as Pierre Simon Laplace (17491827) and Thomas Bayes (17011761). Machine learning is younger, but it has already moved beyond its infancy and can be considered a well-established discipline. Computer science changed our lives several decades ago and continues to do so; but it cannot be considered new.

So, why is data science seen as a novel trend within business reviews, in technology blogs, and at academic conferences?

The novelty of data science is not rooted in the latest scientific knowledge, but in a disruptive change in our society that has been caused by the evolution of technology: datification. Datification is the process of rendering into data aspects of the world that have never been quantified before. At the personal level, the list of datified concepts is very long and still growing: business networks, the lists of books we are reading, the films we enjoy, the food we eat, our physical activity, our purchases, our driving behavior, and so on. Even our thoughts are datified when we publish them on our favorite social network; and in a not so distant future, your gaze could be datified by wearable vision registering devices. At the business level, companies are datifying semi-structured data that were previously discarded: web activity logs, computer network activity, machinery signals, etc. Nonstructured data, such as written reports, e-mails, or voice recordings, are now being stored not only for archive purposes but also to be analyzed.

However, datification is not the only ingredient of the data science revolution. The other ingredient is the democratization of data analysis. Large companies such as Google, Yahoo, IBM, or SAS were the only players in this field when data science had no name. At the beginning of the century, the huge computational resources of those companies allowed them to take advantage of datification by using analytical techniques to develop innovative products and even to take decisions about their own business. Today, the analytical gap between those companies and the rest of the world (companies and people) is shrinking. Access to cloud computing allows any individual to analyze huge amounts of data in short periods of time. Analytical knowledge is free and most of the crucial algorithms that are needed to create a solution can be found, because open-source development is the norm in this field. As a result, the possibility of using rich data to take evidence-based decisions is open to virtually any person or company.

Data science is commonly defined as a methodology by which actionable insights can be inferred from data. This is a subtle but important difference with respect to previous approaches to data analysis, such as business intelligence or exploratory statistics. Performing data science is a task with an ambitious objective: the production of beliefs informed by data and to be used as the basis of decision-making. In the absence of data, beliefs are uninformed and decisions, in the best of cases, are based on best practices or intuition. The representation of complex environments by rich data opens up the possibility of applying all the scientific knowledge we have regarding how to infer knowledge from data.

In general, data science allows us to adopt four different strategies to explore the world using data:

Data science is definitely a cool and trendy discipline that routinely appears in the headlines of very important newspapers and on TV stations. Data scientists are presented in those forums as a scarce and expensive resource. As a result of this situation, data science can be perceived as a complex and scary discipline that is only accessible to a reduced set of geniuses working for major companies. The main purpose of this book is to demystify data science by describing a set of tools and techniques that allows a person with basic skills in computer science, mathematics, and statistics to perform the tasks commonly associated with data science.

To this end, this book has been written under the following assumptions: