Computational Business Analytics

Book Preface

According to the Merriam-Webster dictionary1, analytics is the method of logical analysis. This is a very broad denition of analytics, without an explicitly stated end-goal. A view of analytics within the business community is that analytics describes a process (a method or an analysis) that transforms (hopefully, logically) raw data into actionable knowledge in order to guide strategic decision-making. Along this line, technology research guru Gartner denes analytics as methods that leverage data in a particular functional process (or application) to enable context-specic insight that is actionable (Kirk, 2006). Business analytics naturally concerns the application of analytics in industry, and the title of this book, Computational Business Analytics, refers to the algorithmic process of analytics as implemented via computer.

This book provides a computational account of analytics, and leaves such areas as visualization-based analytics to other authors. Each of the denitions provided above is broad enough to cover any application domain. This book is not intended to cover every possible business vertical, but rather to teach the core tools and techniques applicable across multiple domains. In the process of doing so, we present many examples and a selected number of challenging case studies from interesting domains. Our hope is that practitioners of business analytics will be able to easily see the connections to their own problems and to formulate their own strategies for nding the solutions they seek.

Traditional business analytics has focused mostly on descriptive analyses of structured historical data using myriad statistical techniques. The current trend has been a turn towards predictive analytics and text analytics of unstructured data. Our approach is to augment and enrich numerical statistical techniques with symbolic Articial Intelligence (AI)2 and Machine Learning (ML)3 techniques. Note our usage of the terms augment and enrich as opposed to replace. Traditional statistical approaches are invaluable in datarich environments, but there are areas where AI and ML approaches provide better analyses, especially where there is an abundance of subjective knowledge. Benets of such augmentation include: