1 Introduction
Prescriptive analytics is a part of the business analytics continuum and is also the closest to the decision-making phase. Business analytics is often characterized by three consecutive levels/echelons, i.e., descriptive analytics, predictive analytics, and prescriptive analytics. Prescriptive analytics is the highest level in business analytics and the main focus of this course.
In general terms, business analytics is the art and science of identifying or discovering novel patterns and insights from large volumes and varieties of data utilizing sophisticated machine learning, mathematical, and statistical models to support more accurate and timely managerial decision making (Delen, 2019). Therefore, business analytics is widely perceived as synonymous with managerial decision making and business problem-solving. Figure 1 shows the process of creating information and knowledge through a systematic and scientific transformation of data which leads to making better decisions and, ultimately, achieving “wisdom.”
Figure 1: The process of converting data into knowledge and wisdom
As seen in this figure, various data sources (both structured and unstructured) are converted into mathematical representations (i.e., knowledge models) through a scientific process we now call business analytics.
2 Key Terms
Before we continue, several key terms will provide a foundation for our discussion:
- Business Analytics
- Descriptive analytics
- Predictive analytics
- Prescriptive analytics
- Nature of Data
- Structured data (numeric and non-numeric)
- Unstructured data (textual and multimedia data)
- Operations Research
- Optimization
- Simulation
- Heuristics
- Multi-criteria decision making
3 Analytics and a Simple Analytics Taxonomy
The three commonly-used types/echelons in business analytics (i.e., descriptive, predictive, and prescriptive analytics) provide a structured and comprehensive depiction of the analytics maturity process for organizations (Figure 2). Some sources include diagnostic analytics as a fourth layer/echelon between descriptive and predictive, but most often, this layer of analytics is included as an extension of descriptive analytics. A short description of these analytics echelons, and the questions they are aimed at answering, are given below:
- Descriptive: What happened? Helps to uncover valuable insight into the data being analyzed via the use of on-demand reporting and data visualization.
- Diagnostic: Why did it happen? Helps to identify and understand the causal relationships and related patterns in the data.
- Predictive: What is likely to happen in the future? Helps to forecast the future behavior of people and markets using statistical and machine learning methods.
- Prescriptive: What should I do about it? Now that I know what happened in the past and what might happen in the future, what decisions should I make? Uses operations research methods (optimization, simulation, heuristics, and multi-criteria decision modeling) to provide guidance and understanding on decision alternatives.
Figure 2: The progressive nature of the four types of analytics
4 The Reason behind Analytics Popularity
According to Delen (2021), the reasons behind the popularity of business analytics can be grouped into three main categories, as follows:
- Need for better decisions: Conducting business is no longer “as usual.” Competition has progressively transformed from local to regional, to national, and now, to global. The protections created by tariffs and logistics costs (that sheltered companies into their geographic regions) are no longer as prominent as before. In addition to increased global competition, and perhaps partially because of it, customers have also become more demanding, i.e., asking for the highest quality of products and services, offered at the lowest prices, and delivered quickly. Therefore, data-driven, accurate, optimized, and timely decisions are more critical now than ever before. Analytics offers help with these needs.
- Availability and affordability of enablers: Thanks to recent technological advances and the affordability of software and hardware, organizations are collecting tremendous amounts of data. Internet of Things (IoT)-based data collection systems (which are based on various sensor and RFID [Radio-frequency Identification] technologies, internet, and social media sources) have significantly increased the quantity and quality of organizational data. In addition to the ownership model, cloud-based solutions and software (or hardware) “as-a-service” business models allow small to medium-sized businesses to acquire (i.e., lease and pay only for what they use) analytics capabilities though they have limited financial resources.
- Culture change: There has been a shift from traditional (intuition or “gut-feeling”) decision making to contemporary (fact or evidence-based) decision making at the organizational level. Most successful organizations have made a conscious effort to shift to data- or evidence-driven business practices. Because of the availability of data and supporting information systems infrastructure, such a paradigm shift is taking place faster than many thought possible. As the new generation of quantitatively savvy managers replaces baby boomers, this evidence-based managerial paradigm is expected to accelerate.
5 Prescriptive Analytics and Optimal Decision Making: The Final Frontier
Prescriptive analytics is the highest level or echelon in the analytics hierarchy (Figure 2). It is where the best alternative among many courses of action (which are usually created/identified by descriptive and predictive analytics) are determined using sophisticated mathematical models, often labeled as “operations research” techniques. Therefore, this type of analytics aims to answer the question, “What should I do?” Prescriptive analytics uses well-established operations research techniques like optimization, simulation, and heuristics-based decision modeling. Even though prescriptive analytics is at the top of the analytics hierarchy, its methods are not new. Most of the optimization, simulation, and heuristic techniques that collectively constitute prescriptive analytics were developed during (and right after) World War II in the 1940s and 1950s when there was a dire need to do “the most and best” with limited available resources. Since then, some businesses have used some very specific problem types, including yield/revenue management, transportation modeling, scheduling, etc. The popularity of business analytics and the new taxonomy of analytics have made them popular again, opening their use to a wide array of organizations for various business problems and situations (Delen, 2019).
5.1 Value of Prescriptive Analytics in Business Operations
The value proposition of prescriptive analytics is obvious: supporting optimal decision making. Descriptive and predictive analytics layers focus on information creation, domain understanding, and problem definition. Prescriptive analytics focuses on problem-solving via optimal, timely decision making.
5.2 Suitable Business Decisions for Prescriptive Analytics
Today, most businesses use business analytics to search for novel patterns and correlations, to identify and formulate business problems worthy of solving, and to support optimal and timely managerial decision making. Below are some examples that highlight applications of business analytics across various industries.
- Retail: Retail companies, both online and brick-and-mortar, analyze their customer purchase data to optimize their product offerings, prices, and promotions. The goal is to maximize revenue and profitability while enhancing and maintaining high levels of customer satisfaction and loyalty.
- Finance: Financial institutions use analytics to identify and prevent fraudulent transactions and specifically use predictive and prescriptive analytics to evaluate a person’s financial behavior and assign them a risk level for credit card approval.
- Healthcare: The healthcare industry has identified quite a few ways to use analytics to optimize the allocation of their resources: in the fulfillment of its mission to improve the quality of life for chronically ill patients, provide personalized patient treatment, reduce the rate of hospital-acquired infections, assess and identify treatment risk factors more rapidly, and many more applications. In addition, large medical centers have used free public health data to create visualizations that can help speed up identifying and analyzing healthcare information and tracking the spread of diseases.
- Insurance: The insurance industry has numerous opportunities to expand its use of analytics. For instance, insurance carriers can increase the personalization of services and optimal pricing, which allows for the identification of more specific and actionable consumer segmentation. Additionally, analytics can provide opportunities for better fraud detection and greater industry transparency.
- Transportation: The software Dataiku DSS (Data Science Studio), when applied to freight, sea freight, road freight, and passenger transport, uses predictive and prescriptive analytics with sensor data to determine optimal maintenance schedules.
- Communications, Media, and Entertainment: Entertainment companies such as YouTube, Amazon, and Netflix analyze their users’ browsing habits and patterns, which allows them to create or curate content tailored for specific target audiences optimally.