Overview of Computational Intelligence in Business Analytics: Concepts, Methods, and Tools for Big Data Applications
This chapter is from the book
This chapter is from the book
This chapter is from the book
1.1 Introduction
This book aims to help organizations gain a competitive edge in the marketplace through harnessing the power of computational intelligence approaches. Those approaches—fuzzy sets, artificial neural networks, and genetic algorithms—are at the core of every innovative business, from large corporations to small companies. Businesses that do not leverage computational intelligence will be quickly outperformed by those that do.
The chapters are designed to provide a foundation upon which one can differentiate one’s business through computational intelligence approaches. Thus, the book provides the reader with guidance on how to create acceptable models in a relatively short period of time, and how to arrive at the right innovative decisions before the competition does.
The primary purpose of this book is to facilitate education in the resurgent computational intelligence areas of artificial neural networks, fuzzy sets, and genetic algorithms. The book is written as a text for a course at the graduate or upper-division undergraduate level. It could also be used for short intensive courses of continuing and executive education or as a self-study. No previous knowledge of computational intelligence tools is required to understand the material in this text.
Whereas most (if not all) literature on the topic utilizes statistical software packages, this book urges managers to take advantage of computational intelligence for analysis, exploration, and knowledge generation. As a result, readers are provided with the needed guidance to understand, model, discover, and interpret new patterns and new knowledge from historical evidence and large data sets, and become adept at building powerful models for prediction and classification that do not rely on statistics.
A process based on the exploration of business data with an emphasis on statistical and/or computational intelligence analysis is called Business Analytics. It is used by innovative companies committed to data-driven decision making to gain insights that inform business decisions, and down the road it can be used to automate and optimize business processes. Data obtained through Business Analytics are treated as corporate assets (added value) and are leveraged to gain a competitive edge.
The outcome of Business Analytics depends on data quality (to avoid junk in, junk out), entrepreneurial business analysts who understand the analysis and the business itself, as well as an organizational commitment to data-driven decision making. It should be stressed that entrepreneurial and skillful business analysts are at the core of obtaining competitive advantage for the business. They should operate at every level of their organizations instead of being an elite group of data scientists reporting directly to the executive suite. Those business analysts make the discovered insights actionable, as they discover new knowledge and utilize predictive power of computational intelligence approaches.
Business Analytics itself is used for the following purposes:
- To explore data to find new patterns and relationships (data mining)
- To evaluate and test previous decisions (randomized controlled experiments, multivariate testing)
- To explain why a certain outcome happened (statistical analysis, descriptive analysis)
- To venture into the future (forecast) results (predictive modeling, predictive analytics)
Once the business goal(s) of the analysis is agreed upon, an analysis methodology is selected and data are acquired to support the analysis. Data acquisition often involves extraction from many sources and business systems, data cleaning, dimensionality reduction, feature evaluation, and subsequent integration into a single repository, such as a data mart or a larger data warehouse. Competitive intelligence utilizing intelligent software agents might be used to locate and extract some of the needed data. The analysis is typically performed against a smaller sample set of data to verify its applicability first, and then used on all historical evidence the business has accumulated.
Analytic tools range from spreadsheets with statistical functions to complex data mining and predictive modeling applications. As patterns and relationships in the data are discovered, new questions are asked, new queries are sent, and the analytic process iterates until the business goal is met. This could be optimized by additional technologies such as optimization and/or genetic algorithms.
Deployment of predictive models involves ranking data and information, evaluating their “novelty indexes,” and using the ranks to optimize real-time decisions within applications and business processes. One can also utilize Business Analytics at the tactical level of a business pyramid to tackle unforeseen events, and in many cases the decision making could be automated to support real-time inputs. It is its predictive power that makes computational intelligence insights actionable. The finding associated with an action that is deemed reliable, based upon past data, gives the decision maker a high degree of confidence.
Each chapter in this book is followed by a set of exercises, which are intended to enhance the understanding of the material presented in the chapter. The solutions to a selected subset of these exercises are provided in the instructor’s manual, which also contains further suggestions for use of the text under various circumstances. The exercises are of varying levels of difficulty. The following rating system is applied to approximately indicate the amount of effort required for solving the exercises:
- [Level 1]—Problems at Level 1 are solvable within a day. They test the comprehension and mastery of fundamental concepts. If they relate to the use of software or writing computer code, the programming time is short.
- [Level 2]—Solving problems at Level 2 can take days or weeks (e.g., proof of concept programming or implementation). The chapters provide all the information necessary for solving Level 1 and Level 2 problems.
- [Level 3]—Problems at Level 3 are even harder, and their solutions can take several weeks or even months (e.g., semester-long projects). Many of these exercises are related to innovative avenues of current research.
- [Level 4]—Problems at Level 4 concern open research questions and could be topics of graduate theses or dissertations. Solving Level 3 and Level 4 exercises typically requires doing further literature searches and/or conducting extensive experiments.
It is recommended that the reader do the Level 1 and Level 2 exercises, and tackle at least some of the problems at Levels 3 and 4. Carefully working through Level 1 and Level 2 problems will reward the reader with a thorough understanding of the material of the chapters, and solving Level 3 and Level 4 exercises could turn a reader into an innovator!