Types of Data Analytics

The Analytics Maturity Model

Not all data analysis provides the same level of insight. The four types of analytics are often conceptualized as a progression, moving from simpler, past-focused analysis to more complex, future-focused analysis. This is often called the "Analytics Maturity Model." As we move up the ladder, the analysis becomes more difficult but also provides significantly more value.

The four types are built around answering four simple but powerful questions:

1. Descriptive: What happened?
2. Diagnostic: Why did it happen?
3. Predictive: What is likely to happen?
4. Prescriptive: What should we do about it?

A Clinical Analogy: Understanding the Four Types

Before we dive into formal definitions, let's use a simple analogy from a familiar setting—a doctor's visit—to build your intuition. Given your background in pharmacy, the context of a clinical consultation should be very clear.

The short video below from Maven Analytics uses a great example involving a patient's cholesterol levels to explain the four types in a tangible way. It also makes an important point: these types are not a rigid ladder you must climb and then discard the lower rungs. Instead, they are complementary tools that are often used together.

As the video illustrates:

• Descriptive: "Your cholesterol level is 215." (A factual statement about the past).
• Diagnostic: "This is high, likely due to diet and lack of exercise." (An explanation for why it happened).
• Predictive: "If you continue this lifestyle, your cholesterol will rise, increasing your risk of cardiovascular disease." (A forecast of what will likely happen).
• Prescriptive: "I've prescribed statins and a new diet to lower your cholesterol." (A recommended course of action).

This simple progression from a raw fact to an actionable plan is the essence of the four types of analytics. Now, let's formalize these concepts with definitions and examples from a healthcare context.

The Four Types in Detail

We will now explore each of the four types in more detail, grounding our understanding in authoritative definitions and specific healthcare applications.

The NCBI article on "Healthcare Analytics" provides excellent, concise definitions for each type. We will refer to it as our primary text, supplemented by video explanations from Dr. Chris Paton's lecture on the same topic.

1. Descriptive Analytics: What Happened?

Descriptive analytics is the foundation of all data analysis. It involves summarizing historical data to understand what has already occurred. This is the most common type of analytics, and its goal is to provide a clear, digestible snapshot of the past. It doesn't explain causes or predict the future; it simply presents the facts.

Common outputs include reports, dashboards, and visualizations like bar charts and line graphs showing key metrics, such as the number of prescriptions filled per month or the average patient wait time.

To hear this explained, Dr. Paton provides a quick summary.

Clinical Application: In a pharmacy context, you might use descriptive analytics to create a report showing the top 10 most prescribed medications in your hospital over the last quarter. This report doesn't tell you why those drugs were prescribed most often, but it gives you a clear picture of prescribing patterns.

2. Diagnostic Analytics: Why Did It Happen?

Once you know what happened, the natural next question is why. Diagnostic analytics aims to answer this by drilling down into the data to uncover the root causes of events. It moves beyond simple summaries to explore relationships and correlations in the data. Techniques like root cause analysis, data discovery, and identifying patterns are central to this stage.

Dr. Paton's lecture also provides a concise definition.

Clinical Application: Following our previous example, if descriptive analytics shows a sudden spike in the prescription of a specific antibiotic, diagnostic analytics would be used to investigate why. An analyst might discover a correlation with a recent outbreak of a bacterial infection in a specific hospital ward or find that a new clinical guideline was recently introduced.

3. Predictive Analytics: What Will Happen?

Predictive analytics takes us from looking at the past to forecasting the future. It uses historical data, statistical algorithms, and machine learning techniques to identify the likelihood of future outcomes. This is not a crystal ball; as Dr. Paton notes in his lecture, it provides a probabilistic forecast, not a certainty. Its goal is to provide the best possible estimate of what might happen.

The following video segment provides a fantastic healthcare example of predictive analytics in action: developing an algorithm to predict which newborns are at risk of developing sepsis. This is a powerful illustration of how prediction can lead to proactive care.

Clinical Application: A pharmacy could use predictive analytics to forecast demand for flu vaccines based on historical data, weather patterns, and public health reports. This allows for better inventory management, preventing stockouts during peak season or overstocking that leads to waste.

4. Prescriptive Analytics: What Should We Do?

Prescriptive analytics is the final and most advanced stage. It goes beyond predicting future outcomes to recommend specific actions to take to achieve a desired goal or to mitigate a future risk. It uses a combination of predictive models, algorithms, and business rules to simulate different scenarios and suggest the optimal course of action.

Dr. Paton's lecture also defines this highest level of analytics.

Clinical Application: A prime example in healthcare is a computerized clinical decision support (CDS) system. When a physician enters a prescription for a patient, the system might use prescriptive analytics to cross-reference the patient's record and known drug interactions. If it detects a potential adverse reaction, it will not only flag the issue (predictive) but also suggest an alternative medication (prescriptive).

Summary for a Pharmacy Context

We've covered the four types of analytics with definitions and clinical examples. The image below provides a simple, clean visual summary of the four categories.

To bring this all together specifically for your field, the following table from an article on pharmacy analytics perfectly summarizes the capabilities and strategic benefits of each level within a pharmacy business.

This table powerfully connects the abstract concepts to tangible business and clinical goals that are central to modern pharmacy practice.

Conclusion

In this lesson, we have dissected the broad field of data analytics into its four constituent types. You now have a framework for classifying analytical questions and understanding the progression from hindsight to foresight.

Key Takeaways:

• Descriptive Analytics answers "What happened?" by summarizing historical data.
• Diagnostic Analytics answers "Why did it happen?" by identifying root causes and relationships.
• Predictive Analytics answers "What will likely happen?" by forecasting future outcomes based on historical patterns.
• Prescriptive Analytics answers "What should we do?" by recommending specific actions to achieve a goal or mitigate a risk.
• These four types represent a maturity model, where each successive level offers more value but also requires more complex methods. They are often used together to build a complete picture.

Now that you can distinguish between the different types of analytics, our next lesson will introduce the practical, step-by-step process that analysts follow to answer these questions. We will explore the standard data analysis workflow: Ask, Prepare, Process, Analyze, Share, and Act.