Wrap-Up, Overview, Notes on Implementation, and Current State of Business Analytics
Overview
This book provided an overview of the field of business analytics (BA). BA uses a set of methodology to extract, explore, and analyze big data. It is about extracting information and making decisions from big data. BA is a data-driven decision-making process.
The field of BA can be broken down into two broad areas: (1) business intelligence (BI) and (2) statistical analysis. The flow diagram in Figure 10.1 outlines the broad area of analytics.
Figure 10.1 Broad area of analytics
Chapters 1, 2, and 3 provided an explanation on BI and BA. This book mainly focuses on predictive analytics involving predictive analytics models. Several chapters in the book are devoted to these models.
Broad Areas of Business Analytics
The broad area of BA can be broken down into: (1) BI and (2) statistical analysis
Business Intelligence
BA comes under the broad umbrella of BI discussed in Chapter 3. BI has evolved from business data reporting that involves examining historical data to gain an insight into the performance of a company over time. It involves a number of reporting tools, applications, and methodologies that are used to collect a company’s data (both from internal and external sources) for further analysis, develop queries to get useful information, and create dashboards to aid in data visualization. All this information is used by company executives for data-driven decisions. Figure 10.2 shows the functions of BI and different forms of analytics as applied to different areas.
Figure 10.2 Functions of BI and analytics in different areas
Howard Dresner is credited with first proposing the term BI in 1989, which evolved as the application of data analysis techniques to support business decision making. The tools of BI come from earlier decision support systems (DSSs). A DSS is a collection of applications, algorithms, and computer programs or software designed to analyze and solve specific problems. They use an iterative process that can automate problem solving. These computer-based models are important decision-making tools and can provide a different decision to aid in the decision-making process.
Statistical Analysis
The field of analytics is about driving business decisions using data. Therefore, statistical analysis is at the core of BA. A number of statistical techniques and models—from descriptive and data visualization tools to analytics models—are applied for drawing meaningful conclusions from the data. Statistical analysis involves performing data analysis and creating statistical models and can be broken down into the following categories:
- I.Collect and describe the type of the data to be analyzed.
- II.Explore the relation of the data to the underlying population, (iii) establish and understand how the collected sample data will be used to draw conclusion about the population, (iv) perform data analysis and create different descriptive and predictive analytics models that can be used to predict the future outcomes, and (v) prove the validity of the models.
In its basic form, statistical analysis comprises descriptive statistics and inferential statistics.
Descriptive statistics is the process of describing data using charts and graphs (simple to more advanced). Since companies collect massive amounts of data, the conventional methods of graphical techniques are replaced by software specially designed for big data. The big data software, such as Tableau, is capable of handling massive amounts of data and creating visuals and dashboards that can display the multiple views of business data in one graph.
Inferential statistics is another form of statistics. It is the process of drawing conclusions or making inferences about a population using sample data. A number of inferential statistics tools are used in analytics. Estimation theory, confidence intervals, hypothesis testing, and analysis of variance are a few of the many inferential statistics tools used in BA. Besides the tools of descriptive and inferential statistics, analytics requires an understanding of probability theory and sampling and sampling distributions.
Within the framework of BA and BI, statistical analysis involves inferential statistics that uses sample data to draw conclusion about the population. In statistical analysis, sample is part of a population. Sampling is a systematic way of selecting a few items from the population. The population constitutes the entire data or measurement possible. It is often not possible to study the entire population, so much of the statistical analysis depends on drawing sample from the population. Statistics and statistical analysis, in general, allow us to study the variation in data and allow us to draw inference (conclusion) about a population using sample data. Statistics is the science and mathematics of variation. Statistics is about making decisions from the data.
Statistical analysis also involves identifying trends and patterns in the data to create models that can be used in predictive analytics. These predictive analytics models are used for predicting future business outcomes.
The other component of statistical analysis is data analytics. Statistical analysis and data analytics have somewhat similar approaches, except that data analytics goes beyond statistical analysis that includes more elaborate and extensive applications. The data analytics is explained below.
Data Analytics
Data analytics is the process of exploring and investigating a company’s data to find patterns and relationships in data and applying specialized software. Data analytics makes use of statistical techniques including predictive modeling algorithms to predict business outcomes. These techniques involve testing hypotheses to determine whether hypotheses about a data set are consistent with the stated hypothesis. Data analytics can be thought of as exploratory data analysis (EDA) and confirmatory data analysis that involve the process of data investigation and drawing valid conclusion(s). The term EDA was coined by statistician John W. Tukey in his 1977 book Exploratory Data Analysis.
The terms data analytics, BA, and BI are used interchangeably as the tools used in all these overlap. In a broad sense, data analytics is another approach of exploring and analyzing data using BI, reporting, and online analytical processing (OLAP) as well as BA and advanced analytics tools. Thus, data analytics is also used as an umbrella term that refers to analyzing data and big data with a much broader scope. In the analytics literature, the distinction between data analytics, BI, BA, and advanced analytics is not clear. In some cases, data analytics specifically means data analysis using BA and advanced analytics tools while BI is treated as a separate category. In the earlier chapters of this book, we have tried to explain the tools and applications of each of these terms and tried to outline the differences and similarities among the three—data analytics, BA, and BI. It is important to note that the overall objective of all these tools is to be able to manage, understand, and analyze the massive amounts of data (referred to as big data) using the tools and technologies in making fact-based data-driven business decisions. The tools in all these areas are critical in visualizing data, extracting business trends, studying the current trends, predicting the future business outcomes, optimizing business processes, and using the resources in the most effective way.
All the data analytics and BA initiatives can help businesses increase revenues and profitability, improve operational efficiency, optimize marketing campaigns and respond quickly to changing market trends, improve customer service efforts, address the needs and requirements of customers, gain a competitive edge over the competition, and increase market share. The ultimate goal is to improve business performance by making data-driven decisions. With the advancement in technology, new software and computer applications are being devised to collect and analyze real-time business data, thus enabling real-time analytics.
Types of Data Analytics Applications
Data analytics applications can be categorized into quantitative data analysis and qualitative data analysis. These involve the analysis of qualitative or categorical data and quantitative data. As the name suggests, quantitative analysis involves the analysis of numerical data that are quantifiable and can be compared statistically. The qualitative or categorical data analysis is more interpretive—it focuses on understanding non-numerical data and may involve text, audio and video, images, and interpreting common phrases. A number of applications including text mining and text analytics are now in use to analyze and interpret qualitative data.
Another widely used application of data analytics is BI reporting and OLAP that provides business executives and other corporate workers and stakeholders the data regarding key performance indicators, business operations, and customer information that are valuable decision-making and business process management tools.
One of the major applications of data analytics is the use of data queries to generate reports by BI developers. More recent application is the use of self-service BI tools that enable managers, business analysts, and operations managers to run their own ad hoc queries and build reports themselves.
Business Analytics Models
BA models are divided into:
Descriptive Analytics
Predictive Analytics
Prescriptive Analytics
Advanced Analytics Models
These models are discussed in detail in Chapters 1 and 2. Figure 10.3 outlines the tools and models of BA.
Figure 10.3 Descriptive, predictive, and prescriptive analytics models
The different types of analytics are briefly explained here.
- I.A number of graphical and visual techniques, big data applications, and dashboards are used to visualize a company’s data to learn about the current state of business. This phase is the data visualization part of BA and is known as descriptive analytics.
- II.Information from data visualization is used to model and predict the future business outcomes applying a number of prediction techniques including regression and modeling, time series analysis and forecasting, and data mining techniques to extract useful information from huge databases (known as the knowledge discovery in databases), and, more recently, application of machine learning (ML) and artificial intelligence (AI) techniques is becoming a big part of analytics. This part of analytics is known as predictive analytics.
Most of the descriptive and predictive analytics techniques discussed above use statistical analysis and statistical methods. The predictive analytics techniques mostly use statistical models and algorithms to predict future business trends. These statistical techniques include regression, time series analysis and forecasting, data mining, ML and AI techniques, and also advanced analytics techniques like cluster and classification algorithms in different applications such as marketing analytics. Specific chapters in the book are devoted to these models.
Figures 10.4 and 10.5 summarize models of predictive analytics. An important class of predictive analytics models now includes ML, neural networks, deep learning, and AI. These are shown in Figure 10.5.
Figure 10.4 Predictive analytics models
Figure 10.5 Predictive analytics models including ML, neural networks, deep learning, and AI
The applications of predictive modeling as outlined in Figure 10.5 use DSSs that include expert systems, AI, ML, and deep learning. These are briefly described here.
Artificial Intelligence, Machine Learning, and Deep Learning
AI can be described as the theory and development of computer systems able to perform tasks normally requiring human intelligence, for example, visual perception, speech recognition, language processing, decision making, and translation between languages. First coined in 1956 by John McCarthy, AI involves machines that can perform tasks that are characteristic of human intelligence.
AI systems are classified as (i) weak AI or narrow AI. These systems are designed to perform a narrow task, the phenomenon that machines are not too intelligent to do their own work. In this type of AI system, the system works based on the information and rules fed to the system. Apple’s SIRI is an example. Google’s AI system is also an example of a narrow AI that is stronger than SIRI. Another example of this category is a poker game where a machine can play and beat humans based on all rules and moves fed into the machine. (ii) Strong AI systems are the machines that can actually think and perform tasks on its own like human beings. There is no existing application that can be described as a true strong AI. This is an active area of research where these systems are evolving rapidly and getting close to building strong AI systems.
Machine Learning
AI and ML are sometimes used synonymously, but there is a difference between the two. ML is simply a way of achieving AI.
AI can be achieved without using ML, in which the AI system would require specific program with millions of lines of codes with complex rules and decision trees. Alternatively, ML algorithms can be developed. These are a way of “training” an algorithm so that it can learn how. The “training” requires feeding huge amounts of data to the algorithm and allowing it to adjust, learn, and improve. One of the most successful applications of ML is in the area of computer vision—the ability of a machine to recognize an object in an image or video.
Deep Learning
Deep learning is a class of ML algorithm and is one of many approaches to ML. Most deep learning models are based on an artificial neural network and are inspired by the structure and function of the brain or neurons in the brain. The deep learning applications are commonly referred to as artificial neural networks (ANNs). The term deep refers to the number of layers through which the data are transformed. The reported applications of deep learning include computer vision, speech recognition, natural language processing, social network filtering, bioinformatics, drug design, medical image processing, material inspection, and more. The research in this area is promising, and the results produced in different applications are comparable to and, in some cases, superior to human experts.9
Background and Prerequisites to Predictive Analytics
The application and implementation of predictive analytics models require the necessary background and understanding of several of the statistical concepts. These include probability and probability distributions, estimation theory and confidence intervals, sampling theory, hypothesis testing, and covariance analysis. These are shown in Figure 10.6 and are explained in the Appendix. The readers may refer to Appendix A through D for the explanation of these topics.
Figure 10.6 Background and prerequisites to predictive analytics
Optimization Models for Business Analytics Prescriptive Analytics
BA also involves using a number of optimization, simulation, operations management, and business process optimization techniques to optimize business performances. This is the prescriptive analytics phase of BA. The models under prescriptive analytics are also known as advanced analytics models. Figure 10.7 shows the most common optimization and other models. This text mainly focuses on predictive analytics; therefore, prescriptive models are not discussed in detail.
Figure 10.7 Prescriptive analytics models
In this chapter, we have provided an overview. Separate chapters in the book discuss these concepts in detail. The first three chapters of the book are devoted to the basic concepts and models used in analytics, BA, and BI. In Chapter 4, we discussed the descriptive analytics along with applications and a case. We explained the objectives of descriptive analytics and how it leads to predictive analytics. Chapter 5 draws a distinction between descriptive and predictive analytics and the background and prerequisites needed to apply predictive analytics models. These include probability and probability distributions, sampling and sampling distribution, tools of inferential statistics—estimation and confidence intervals, hypothesis testing, and correlation analysis. The applications, examples, and importance of all these concepts are presented in this chapter. A detailed treatment of all these concepts is provided in Appendixes A–D. The appendixes are available for a free download to the readers.
Chapter 6 provides an overview of the most widely used predictive analytics models. Each model is discussed along with their purpose, tools, and applications. The brief explanation in this chapter of each model provides the reader the importance and purpose behind each model to follow.
In Chapters 7 and 8, we discussed the most widely used predictive analytics model in detail. These include regression analysis and different regression models with applications and examples. Chapter 8 provides another class of predictive models—time series analysis and forecasting. Chapters 7 and 8 contain a number of different models with examples. These are very widely used models in predictive analytics. Chapter 9 provides an introduction to data mining—an important part of data analytics, data analysis, and predictive modeling. The purpose and importance of data mining in predictive modeling are explained. We also introduce the recent applications of ML, AI, deep learning, and neural networks models. These are now an integral part of predictive analytics and are finding applications in a number of areas. Finally, the prescriptive analytics is introduced along with models they use and the purpose of prescriptive analytics in the overall BA.
Future of Data Analytics and Business Analytics
Job Outlook
- •Demand for skilled data scientists continues to be sky-high, with IBM recently predicting that there will be a 28% increase in the number of data scientists employed in the next 2 years.
- •According to the U.S. Bureau of Labor Statistics (BLS) data, employment of management analysts—including business analysts—is expected to grow 14 percent from 2014 to 2024, which is much faster than the average for all occupations.
- •The BLS reports for May 2016 showed that the average annual income for all management analysts, including business analysts, was $91,910. The middle 50 percent earned between $60,950 and $109,170. Salaries for the lowest 10 percent were around $46,560, while the highest 10 percent brought in upward of $149,720.
Here are some other facts:
- •The amount of data doubles every 3 years as various digital sources continue to make information available (Source: McKinsey & Company).
- •A significant shortage of managers and analysts who can effectively use big data analytics and analytical concepts to make decisions is predicted for 2018 (Source: McKinsey & Company).
- •Three-quarters of companies are missing the skills and technology to make the best use of the data they collect (Source: PWC).
- •Businesses in all industries are beginning to capitalize on the vast increase in data and the new big data technologies becoming available for analyzing and gaining value from it. This makes it a great prospect for anyone looking for a well-paid career in an exciting and cutting-edge field.
- •Analytics, ML, AI, and expert systems applications and research are not for just those following a traditional academic path. A number of industries including Google, Amazon, IBM, and others are highly invested in big data analytics, AI, ML, and deep leaning research and applications.
Certification and Online Courses in Business Analytics
There are also a large number of free online courses and tutorials which a motivated individual could use as a springboard into a rewarding and lucrative career (please follow the link below).
Foundations in Business Analytics — University of Maryland
Business Analytics Certificate — Cornell University
Master Certificate in Business Aanalytics — Michigan State University
The above are listed as the Best Online Business Analytics Certificates & Courses [Updated 2018]
Summary
In this chapter, we provided an overview of the field of analytics. The broad area of analytics can be divided into two broad categories: BI and statistical analysis.
BI evolved as the application of data analysis techniques to support business decision-making processes. The tools of BI come from earlier DSSs. These computer-based models are important decision-making tools and can provide different alternatives to aid in the decision-making process. The second broad area of analytics is statistical analysis. The field of analytics is about driving business decisions using data. Therefore, statistical analysis is at the core of BA. A number of statistical techniques and models—from descriptive and data visualization tools to analytics models—are applied for drawing meaningful conclusions from the data.
Statistical analysis involves performing data analysis and creating statistical models and can be broken down into (i) data analytics, (ii) BA, and (iii) advanced analytics.
Data analytics is the process of exploring and investigating a company’s data to find patterns and relationships in data and applying specialized software to learn the current state of business through data visualization, predict future business outcomes, and optimizing business process. Thus, data analytics is exploring and analyzing data using BI, reporting, and OLAP as well as BA and advanced analytics tools. Data analytics is also used as an umbrella term that refers to analyzing data and big data with a much broader scope.
The BA area is divided into: (i) descriptive analytics, (ii) predictive analytics, (iii) prescriptive analytics, and (iv) advanced analytics models. These involve a number of models and tools that were briefly described throughout the book.
Most of the descriptive and predictive analytics techniques use statistical analysis and statistical methods. The focus of this book is predictive analytics. Predictive analytics techniques use mostly statistical models and algorithms to predict future business trends. These statistical techniques include regression, time series analysis and forecasting, data mining, ML, and AI techniques and also advanced analytics techniques like cluster and classification algorithms in different applications such as marketing analytics. Specific chapters in the book are devoted to these models. These are described in different chapters of the book.
BA initiatives can help businesses increase revenues and profitability, improve operational efficiency, optimize marketing campaigns and respond quickly changing market trends, improve customer service efforts, address the needs and requirements of customers, gain a competitive edge over the competition, and increase market share. The ultimate goal is to improve the business performance by making data-driven decisions.
Finally, we explored the future and job outlook of this emerging field of BA. This is one of the fastest growing areas that is predicted to have immense job growth and opportunities. According to the BLS, employment of management analysts—including business analysts—is expected to grow 14 percent from 2014 to 2024, which is much faster than the average for all occupations. A number of universities and agencies are now offering courses, graduate degrees, and certifications in BA. The opportunities were listed in this chapter.