Business Analytics: New Frontiers?
The availability of data and affordability of technological enablers has made business analytics a buzzword in business circles. So, what is business analytics?
business analytics refers to the tools and techniques used to convert raw data into actionable insight for better decision making
Begin with the novel ways in which data is now captured with the widespread adoption of the Internet. These include:
RFID tags
Digital energy meters
Clickstream Weblogs
Smart home devices
Wearable health devices
Social network data
Analysis of unstructured data, however, poses challenges to conventional computational systems, from both software as well as hardware perspectives. To deal with these challenges hardware (e.g. parallel processing with computational memory and multi-processor computing systems) as well as software (e.g. Hadoop and NoSQL) have been developed to address the challenges of “Big Data.”
Analytics can broadly be conceptualised into three hierarchical but overlapping groupings. They are descriptive/diagnostic analytics, predictive analytics, and prescriptive analytics. Most organisations start with descriptive analytics, proceed to predictive analytics, before reaching for the apex of the analytics hierarchy—prescriptive analytics. Even though moving between each level is not separable, there is a distinct difference in complexity between the groupings. A business may primarily be engaged with descriptive analytics while at the same time using predictive capabilities in a piecemeal fashion.
Descriptive analytics
Descriptive analytics serves as an entry point into the world of analytics. Also referred to as business intelligence, most of the activities at this level deal with presenting reports to summarise past business activity, to answer the question of “What happened?” The objective of descriptive analytics is to articulate well-defined business problems and opportunities.
Business reports can take shape in the form of static snapshots of transactions delivered to decision-makers on a regular schedule (e.g., daily, weekly, quarterly); dynamic views of business performance indicators delivered to managers and executives in a digestible form on a continuous basis; or ad-hoc reports where the decision-maker has the capability of creating customised reports using an intuitive, drag-and-drop graphical user interface (GUI) to address a particular situation. Tableau and Power BI dominate this space with industry-leading products that exploit information visualisation theories, to empower users with the insight needed to make evidence-based decisions at scale.
Diagnostic analytics as a logical extension scrutinises data further to answer the question of “why did it happen?” It encourages exploratory data analysis (EDA) using techniques like drill-down, data discovery, and data mining to identify root causes.
Predictive Analytics
Organisations that have matured move into predictive analytics, where they look beyond what happened and try to answer the question of “What will happen?” The objective of predictive analytics is to make accurate projections of future events and outcomes.
Prediction is the process of formulating estimates about the future value of some variable like customer demand, interest rates, stock market movements, etc. If what is being predicted is a categorical variable, the act of prediction is called classification; otherwise, it is referred to as regression. If the predicted variable is time-dependent, then the prediction process is labelled time-series forecasting. Popular machine learning models that learn from historical data to predict future events include linear and logistic regression, as well as neural networks, support vector machines, decision trees, and others. A mixture of open-source software (R and Python), GUIs (Alteryx and Tableau with Einstein Analytics), and cloud computing infrastructure (Amazon Web Services and Microsoft Azure) currently offer ‘off-the-shelf’ production-ready algorithms.
Despite the fervour, predictive analytics is limited in application by the fact that it relies on making educated guesses about the future using historical data. For predictions to be accurate, the past must resemble the future. While this is a reasonable assumption in the main, navigating black swans events (or unforeseen consequential events) require different approaches.
Prescriptive Analytics
Prescriptive analytics sits at the upper echelon of the analytics hierarchy. It is where the best alternative is determined using complex mathematical models. In a sense, prescriptive analytics tries to answer the question of “What should I do?” The objective is to make the best decision among the alternatives.
Prescriptive analytics utilises optimisation, simulation, and heuristics-based decision modelling techniques for specific problem types, such as yield and revenue management, transportation modelling and scheduling. Most of the optimisation and simulation models that constitute prescriptive analytics today were developed in the 1940s when the need for optimal resource allocation was dire and has been applied extensively to industries as varied as transportation and logistics to healthcare. Excel Add-ins (Solver and @RISK) and specialist software (Simul8) offer solutions to optimisation and simulation type problems, respectively.
Business journals and magazines are saturated with articles on analytics because it is upending the way managerial decisions are being made. Success, however, depends largely on the volume and representational richness of the data; its accuracy and timeliness; and the capabilities and sophistication of the analytical tools and procedures used in the process. Understanding the analytics topography can help in the selection and implementation of the capabilities required to navigate a hyper-competitive global environment successfully.
References
Reinhart, C. M., & Rogoff, K. S. (2011). This Time Is Different: Eight Centuries of Financial Folly (Reprint ed.). Princeton University Press.
Taleb, N. N. (2010). The Black Swan: Second Edition: The Impact of the Highly Improbable: With a new section: “On Robustness and Fragility” (Incerto) (2nd ed.). Random House Trade Paperbacks.