There are four primary types of analytics that everyone working with data should know. Even though they all fall within the umbrella of Data Analytics, they have entirely different objectives. It’s important to understand those distinctions if you’re going to use data appropriately and effectively. So, buckle up, and let’s dive in.

Descriptive Analytics (How/When/What Happened?)

Descriptive Analytics is the foundation of all the other types of analytics. It seeks to answer the questions, “how/when/what happened?” Since it is a hindsight analysis, it looks at past measurements and is renders its findings after the fact. The utility of Descriptive Analysis therefore often depends on the length of time between data collection and data description.

The majority of analytics that you encounter are Descriptive Analytics. You see these all the time. Profit and loss statements, survey result summaries, and social media follower reports are all common examples. If you’re answering questions like, “what is the average customer lifetime value?” or “What post is getting the most engagement in North America?”, you are performing Descriptive Analytics.

Some standard techniques you’ll use in Descriptive Analytics include data aggregation, summary statistics, and clustering

Descriptive Analytics Never Tell You Why Something Happened

Descriptive Analytics only tells you how/when/what happened. It never tells you why something happened. Make sure you and everyone on your team using that data understands that reality. Be careful to steer your discussions away from any diagnosis. For that, you’ll need to enter the next domain of analytics: Diagnostic Analytics. 

Diagnostic Analytics (Why Did It Happen?)

Diagnostic Analytics is the first part of the juicy analytics core that everyone wants to taste. It sits just outside of Descriptive Analytics and attempts to answer the question, “Why did it happen?” 

Diagnostic Analytics is still a hindsight analysis. It looks backward in an attempt to understand the mechanics behind observations. It searches for and explores patterns in the data and attempts to identify relationships. 

Techniques used at this stage include sensitivity analysisprincipal component analysis, and regression

The Role of Diagnostic Analytics in Experimentation

In it’s more advanced states, Diagnostic Analytics is central to experimentation. Experimentation is the only way you can establish causation. Scientists will review the findings from early Diagnostic Analysis, form hypotheses, and design experiments. If the results of those experiments align with a body of evidence, Diagnostic Analysis can establish a high probability of causation.

In most business cases, however, there are inadequate time and resources to perform experiments and await the results. Therefore, the strength of correlation analysis in this phase dictates how the business responds to its data.

Predictive Analytics (What is Going to Happen?)

Predictive Analytics is where things start to get really interesting. The previous types of analytics described what has happened and established diagnostic correlations. Once you have those ingredients, you can gaze into your statistical crystal ball and predict the future.

The goal of Predictive Analytics is forecasting. It evaluates the probability of something happening under a specific set of conditions. As you can imagine, this is a highly technical form of analytics. Staff outside of your organization’s data analytics team should not attempt it. Additionally, it should always be supervised by an experienced senior analyst or data scientist.

Some standard methods used in Predictive Analytics include machine learning algorithmsclassification, and regression models

In the right hands, Predictive Analytics is a potent instrument and it is almost always worth the investment in terms of time and resources.

Prescriptive Analytics (Cause Something to Happen)

Very few organizations make it this far along the analytics spectrum. Just like a doctor writing a prescription to treat an illness, Prescriptive Analytics makes nuanced recommendations to induce some desired effect.

To unlock this dark magic, advanced analysts use techniques like artificial intelligencemachine learning, and neural networks to evaluate parameters and provide targeted suggestions. When done correctly, an organization’s leaders can translate these prescriptions into goal-oriented strategies.

The keyword here is specific. Prescriptive analytics is most reliable when the focus is narrow, and the operating conditions for the goal are known. Organizations that apply their prescriptions to too broad a scope risk making inaccurate assumptions and subsequently crafting poor strategy.

All Four Types of Analytics can be Within Reach of Most Organizations

All four of these types of analytics are within reach of most organizations, despite the progressively complex prerequisites. The first step is to foster a culture of data literacy at your organization. You won’t be able even to make it past Descriptive Analytics if your organization is data illiterate. 

The second step is to find the right expertise. Ideally, this is in-house expertise. An in-house analytics team will have a better sense of what is important to your business. It will also have a grasp on how your business works. If that pathway is not an option, and you might consider hiring a data firm to help. Be careful, however, of firms that offer “fully automated analytics.” You want a dedicated team of analysts working on your predictions and prescriptions.

Advertisements Disclosure

I will always make it clear if I am writing to endorse or recommend a specific product(s) or service(s). I hate it when I visit a site only to find out that the article is just one big ad.

Various ads may be displayed on this post to help defray the operating cost of this blog. I may make a small commission on any purchases you make by clicking on those advertisements. Thank you for supporting my work bringing you accurate and actionable information on data literacy, analytics, and engineering.