Analytics as a Functional Domain

To analyse is to examine data carefully and in detail so as to identify causes, key factors, possible results, etc. ‘Analytics’ is the process of analysis of data that is done logically aided by sciences (statistical, computers, etc.). ‘Business Analytics’ is a term that can be defined as “a set of all the skills, technologies, applications and practices required for continuous iterative exploration and investigation of past business performance to gain insight and drive business planning (Beller & Barnett, 2009). This process, depending upon its outcomes, can be descriptive, diagnostic, predictive, or prescriptive in nature (Gartner Report, 2012) (Figure 1). Some Illustrative Analytical Approaches

• Descriptive analytics describes a phenomenon through different measures that could capture its relevant dimensions. The purpose is to simply unravel ‘what happened’ or alerting on what is going to happen.
• Diagnostic analytics evaluates ‘why’ something happened. It needs exploratory data analysis of the existing data or additional data if required to be collected using tools such as visualization techniques in order to discover the root causes of a problem.
• Predictive analytics seeks options for future business imperatives, predicts potential future outcomes, and explains drivers of the observed phenomena using statistical or data mining techniques – for instance, predicting the sales of a product for the next month or the behaviour of a target segment of the customers.
• Prescriptive analytics goes beyond describing, explaining, and predicting to suggest ‘what courses of action may be taken’ for the future to optimize business processes in order to achieve business objectives. In other words, it associates decision alternatives with the prediction of outcomes. For prescriptive analytics, decision analysis is used which includes tools such as optimization and simulation.

In a typical organization, analytics could be used in different forms: a) It could be a dashboard kind of application when an organization routinely generates various metrics using data to monitor a process or multiple processes across time. All decision support systems fall into this category. This kind of application could be meaningful to understand, say, the financial health of an organization at a given point of time or to compare it with others or its own across different points of time. Even in such routine applications, what is important is to interpret these numbers and meaningfully connect it with the understanding of the underlying process relevant to decision-making. For such analytics applications, one needs to cultivate the skill of reading relevant facts from figures, connecting them with the relevant decision-making process, and finally, taking a data-driven decision from the business point of view. We would categorize such applications as “descriptive” analytics. Usually, such analytics is used repetitively and routinely in an organization for its day-to-day operations. b) The second kind of analytics falls under the application category which is not routinely used in an organization. These are investigative in nature and could be either exploratory or confirmatory. For such applications, organizations usually hire consultants or sleuths. Given a business objective, the tasks of the sleuths could be to frame relevant research questions, collect appropriate data, and analyse it intelligently, and finally, to connect the findings with the business objective. Often, during the data collection or analysis stage, some interesting patterns emerge that are considered to be surprising discoveries. For this kind of application, one needs to be smart in posing relevant questions, obtaining intelligent answers with appropriate data, possessing skills of exploring data with the mind of a sleuth and finally connecting the findings with the business objective. Such applications are useful in understanding the business environment, the customers, the risks associated with a new product, etc., generally in making strategic decisions. Such analytics are generally used for forward-looking decision- making. We could categorize these attempts as “diagnostic” or, in some instances, where there are direct implications on the future, as “predictive” analytics. c) Unfortunately, there are very few examples of good “prescriptive” analytics in the real world. A good reason for the shortfall is that most databases are constrained on the number of dimensions that they capture. Hence the analysis from such data provides, at best, partial insights into a complex business problem. Most prescriptive analytics exercises are therefore half-baked and need to be used with caution. Nevertheless, business analysts have devised “scenario builders” based on statistical analysis of market response data which provide elasticity measures (impacts) of different managerially controlled parameters. Using them, they have devised “what if” simulators that help provide insights about what may be the plausible options that the business ought to implement in order to maintain or strengthen its position in the market. d) There is another “category” of analytics — Big Data Analytics – which appears to be more of a buzz word in today’s analytics parlance. Today’s companies process more than 60 terabytes of data annually which is 1,000 times more than what they used to do a decade Also, huge amounts of data are generated by individuals spread across different geographic regions in different formats like texts, videos, tweets, blogs, etc. More importantly, in 1986, only six percent of the world’s data were in digital format compared to 90 percent of it today. So, the real concern is to make use of this huge volume of data to derive meaningful insights and drive fact-based decisions for business success. In the world of big data, filtering the signal out of the noise is the key. Mostly the analytics is exploratory (descriptive) in nature. By making use of exploratory statistical methods (data mining tools), the sole objective is to discover meaningful patterns or unknown correlations that could be used for making business decisions. A later section provides a more detailed perspective about the potential of this emerging field. In a nutshell, business analytics supposedly spans the past, present, and future to give us more knowledge, better information, and concrete insights. It tends to get more complex and valuable as it moves from descriptive to prescriptive applications.   Republished on authors consent from Vikalpa